Notebooks/CNN+LSTM_Violence_Detection.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Copy of Violence-Detection.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "PoL9g34z3g4y",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "9b335640-0999-4de3-8157-b6554c943cb2"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "r33tE-Pb9pv6"
      },
      "source": [
        "%matplotlib inline\n",
        "import cv2\n",
        "import os\n",
        "import numpy as np\n",
        "import keras\n",
        "import matplotlib.pyplot as plt\n",
        "# import download\n",
        "from random import shuffle\n",
        "from keras.applications import VGG16\n",
        "from keras import backend as K\n",
        "from keras.models import Model, Sequential\n",
        "from keras.layers import Input\n",
        "from keras.layers import LSTM\n",
        "from keras.layers import Dense, Activation\n",
        "import sys\n",
        "import h5py"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ENPwak-qlfMC"
      },
      "source": [
        "count=0;\n",
        "import cv2"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "poisr8h4-QVV",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "0f4d7c01-929c-4da4-ba17-91c4d879001b"
      },
      "source": [
        "keras.__version__"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'2.4.3'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lcs3FECO-RU-"
      },
      "source": [
        "def print_progress(count, max_count):\n",
        "    # Percentage completion.\n",
        "    pct_complete = count / max_count\n",
        "\n",
        "    # Status-message. Note the \\r which means the line should\n",
        "    # overwrite itself.\n",
        "    msg = \"\\r- Progress: {0:.1%}\".format(pct_complete)\n",
        "\n",
        "    # Print it.\n",
        "    sys.stdout.write(msg)\n",
        "    sys.stdout.flush()\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Yj2IInCllO7a"
      },
      "source": [
        "for name in range(1,6):\n",
        "  for filename in os.listdir('drive/My Drive/CMP-258 Project/dataset/movies/'+str(name)+'/Violence'):\n",
        "    vidObj = cv2.VideoCapture('drive/My Drive/CMP-258 Project/dataset/movies/'+str(name)+'/Violence/'+filename) \n",
        "    while True: \n",
        "          success, image = vidObj.read() \n",
        "          #print(success)\n",
        "          # Saves the frames with frame-count \n",
        "          if success:\n",
        "            cv2.resize(image,(240,240));\n",
        "            cv2.imwrite(\"drive/My Drive/CMP-258 Project/dataset/DatasetFrames/Fight/frame%d.jpg\" % count, image) \n",
        "            count += 1\n",
        "          else:\n",
        "            break\n",
        "    vidObj.release() \n",
        "    cv2.destroyAllWindows() "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VcJA116spOhs",
        "outputId": "d1bb2f8d-728e-4ff3-8507-8dbbbfd23d4e"
      },
      "source": [
        "count"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "12530"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sLRTdsJMpQva"
      },
      "source": [
        "count=0"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RgW38ldJpSYE"
      },
      "source": [
        "for name in range(1,6):\n",
        "  for filename in os.listdir('drive/My Drive/CMP-258 Project/dataset/movies/'+str(name)+'/NonViolence'):\n",
        "    # print(filename)\n",
        "    vidObj = cv2.VideoCapture('drive/My Drive/CMP-258 Project/dataset/movies/'+str(name)+'/NonViolence/'+filename) \n",
        "    while True: \n",
        "          success, image = vidObj.read() \n",
        "          #print(success)\n",
        "          # Saves the frames with frame-count \n",
        "          if success:\n",
        "            cv2.resize(image,(240,240));\n",
        "            cv2.imwrite(\"drive/My Drive/CMP-258 Project/dataset/DatasetFrames/NoFight/frame%d.jpg\" % count, image) \n",
        "            count += 1\n",
        "          else:\n",
        "            break\n",
        "    vidObj.release() \n",
        "    cv2.destroyAllWindows() "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZA5nmDNE4UQq",
        "outputId": "864310ca-bca3-4c13-953e-95ab83ca94fb"
      },
      "source": [
        "count"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "9544"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uRPHF2R55QB4",
        "outputId": "aee8af3a-9760-4038-d0dd-fce77e8af29b"
      },
      "source": [
        "!unzip \"drive/MyDrive/CMP-258 Project/archive.zip\""
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Archive:  drive/MyDrive/CMP-258 Project/archive.zip\n",
            "  inflating: data/fi100_xvid.avi     \n",
            "  inflating: data/fi101_xvid.avi     \n",
            "  inflating: data/fi102_xvid.avi     \n",
            "  inflating: data/fi103_xvid.avi     \n",
            "  inflating: data/fi104_xvid.avi     \n",
            "  inflating: data/fi105_xvid.avi     \n",
            "  inflating: data/fi106_xvid.avi     \n",
            "  inflating: data/fi107_xvid.avi     \n",
            "  inflating: data/fi108_xvid.avi     \n",
            "  inflating: data/fi109_xvid.avi     \n",
            "  inflating: data/fi10_xvid.avi      \n",
            "  inflating: data/fi110_xvid.avi     \n",
            "  inflating: data/fi111_xvid.avi     \n",
            "  inflating: data/fi112_xvid.avi     \n",
            "  inflating: data/fi113_xvid.avi     \n",
            "  inflating: data/fi114_xvid.avi     \n",
            "  inflating: data/fi115_xvid.avi     \n",
            "  inflating: data/fi116_xvid.avi     \n",
            "  inflating: data/fi117_xvid.avi     \n",
            "  inflating: data/fi118_xvid.avi     \n",
            "  inflating: data/fi119_xvid.avi     \n",
            "  inflating: data/fi11_xvid.avi      \n",
            "  inflating: data/fi120_xvid.avi     \n",
            "  inflating: data/fi121_xvid.avi     \n",
            "  inflating: data/fi122_xvid.avi     \n",
            "  inflating: data/fi123_xvid.avi     \n",
            "  inflating: data/fi124_xvid.avi     \n",
            "  inflating: data/fi125_xvid.avi     \n",
            "  inflating: data/fi126_xvid.avi     \n",
            "  inflating: data/fi127_xvid.avi     \n",
            "  inflating: data/fi128_xvid.avi     \n",
            "  inflating: data/fi129_xvid.avi     \n",
            "  inflating: data/fi12_xvid.avi      \n",
            "  inflating: data/fi130_xvid.avi     \n",
            "  inflating: data/fi131_xvid.avi     \n",
            "  inflating: data/fi132_xvid.avi     \n",
            "  inflating: data/fi133_xvid.avi     \n",
            "  inflating: data/fi134_xvid.avi     \n",
            "  inflating: data/fi135_xvid.avi     \n",
            "  inflating: data/fi136_xvid.avi     \n",
            "  inflating: data/fi137_xvid.avi     \n",
            "  inflating: data/fi138_xvid.avi     \n",
            "  inflating: data/fi139_xvid.avi     \n",
            "  inflating: data/fi13_xvid.avi      \n",
            "  inflating: data/fi140_xvid.avi     \n",
            "  inflating: data/fi141_xvid.avi     \n",
            "  inflating: data/fi142_xvid.avi     \n",
            "  inflating: data/fi143_xvid.avi     \n",
            "  inflating: data/fi144_xvid.avi     \n",
            "  inflating: data/fi145_xvid.avi     \n",
            "  inflating: data/fi146_xvid.avi     \n",
            "  inflating: data/fi147_xvid.avi     \n",
            "  inflating: data/fi148_xvid.avi     \n",
            "  inflating: data/fi149_xvid.avi     \n",
            "  inflating: data/fi14_xvid.avi      \n",
            "  inflating: data/fi150_xvid.avi     \n",
            "  inflating: data/fi151_xvid.avi     \n",
            "  inflating: data/fi152_xvid.avi     \n",
            "  inflating: data/fi153_xvid.avi     \n",
            "  inflating: data/fi154_xvid.avi     \n",
            "  inflating: data/fi155_xvid.avi     \n",
            "  inflating: data/fi156_xvid.avi     \n",
            "  inflating: data/fi157_xvid.avi     \n",
            "  inflating: data/fi158_xvid.avi     \n",
            "  inflating: data/fi159_xvid.avi     \n",
            "  inflating: data/fi15_xvid.avi      \n",
            "  inflating: data/fi160_xvid.avi     \n",
            "  inflating: data/fi161_xvid.avi     \n",
            "  inflating: data/fi162_xvid.avi     \n",
            "  inflating: data/fi163_xvid.avi     \n",
            "  inflating: data/fi164_xvid.avi     \n",
            "  inflating: data/fi165_xvid.avi     \n",
            "  inflating: data/fi166_xvid.avi     \n",
            "  inflating: data/fi167_xvid.avi     \n",
            "  inflating: data/fi168_xvid.avi     \n",
            "  inflating: data/fi169_xvid.avi     \n",
            "  inflating: data/fi16_xvid.avi      \n",
            "  inflating: data/fi170_xvid.avi     \n",
            "  inflating: data/fi171_xvid.avi     \n",
            "  inflating: data/fi172_xvid.avi     \n",
            "  inflating: data/fi173_xvid.avi     \n",
            "  inflating: data/fi174_xvid.avi     \n",
            "  inflating: data/fi175_xvid.avi     \n",
            "  inflating: data/fi176_xvid.avi     \n",
            "  inflating: data/fi177_xvid.avi     \n",
            "  inflating: data/fi178_xvid.avi     \n",
            "  inflating: data/fi179_xvid.avi     \n",
            "  inflating: data/fi17_xvid.avi      \n",
            "  inflating: data/fi180_xvid.avi     \n",
            "  inflating: data/fi181_xvid.avi     \n",
            "  inflating: data/fi182_xvid.avi     \n",
            "  inflating: data/fi183_xvid.avi     \n",
            "  inflating: data/fi184_xvid.avi     \n",
            "  inflating: data/fi185_xvid.avi     \n",
            "  inflating: data/fi186_xvid.avi     \n",
            "  inflating: data/fi187_xvid.avi     \n",
            "  inflating: data/fi188_xvid.avi     \n",
            "  inflating: data/fi189_xvid.avi     \n",
            "  inflating: data/fi18_xvid.avi      \n",
            "  inflating: data/fi190_xvid.avi     \n",
            "  inflating: data/fi191_xvid.avi     \n",
            "  inflating: data/fi192_xvid.avi     \n",
            "  inflating: data/fi193_xvid.avi     \n",
            "  inflating: data/fi194_xvid.avi     \n",
            "  inflating: data/fi195_xvid.avi     \n",
            "  inflating: data/fi196_xvid.avi     \n",
            "  inflating: data/fi197_xvid.avi     \n",
            "  inflating: data/fi198_xvid.avi     \n",
            "  inflating: data/fi199_xvid.avi     \n",
            "  inflating: data/fi19_xvid.avi      \n",
            "  inflating: data/fi1_xvid.avi       \n",
            "  inflating: data/fi200_xvid.avi     \n",
            "  inflating: data/fi201_xvid.avi     \n",
            "  inflating: data/fi202_xvid.avi     \n",
            "  inflating: data/fi203_xvid.avi     \n",
            "  inflating: data/fi204_xvid.avi     \n",
            "  inflating: data/fi205_xvid.avi     \n",
            "  inflating: data/fi206_xvid.avi     \n",
            "  inflating: data/fi207_xvid.avi     \n",
            "  inflating: data/fi208_xvid.avi     \n",
            "  inflating: data/fi209_xvid.avi     \n",
            "  inflating: data/fi20_xvid.avi      \n",
            "  inflating: data/fi210_xvid.avi     \n",
            "  inflating: data/fi211_xvid.avi     \n",
            "  inflating: data/fi212_xvid.avi     \n",
            "  inflating: data/fi213_xvid.avi     \n",
            "  inflating: data/fi214_xvid.avi     \n",
            "  inflating: data/fi215_xvid.avi     \n",
            "  inflating: data/fi216_xvid.avi     \n",
            "  inflating: data/fi217_xvid.avi     \n",
            "  inflating: data/fi218_xvid.avi     \n",
            "  inflating: data/fi219_xvid.avi     \n",
            "  inflating: data/fi21_xvid.avi      \n",
            "  inflating: data/fi220_xvid.avi     \n",
            "  inflating: data/fi221_xvid.avi     \n",
            "  inflating: data/fi222_xvid.avi     \n",
            "  inflating: data/fi223_xvid.avi     \n",
            "  inflating: data/fi224_xvid.avi     \n",
            "  inflating: data/fi225_xvid.avi     \n",
            "  inflating: data/fi226_xvid.avi     \n",
            "  inflating: data/fi227_xvid.avi     \n",
            "  inflating: data/fi228_xvid.avi     \n",
            "  inflating: data/fi229_xvid.avi     \n",
            "  inflating: data/fi22_xvid.avi      \n",
            "  inflating: data/fi230_xvid.avi     \n",
            "  inflating: data/fi231_xvid.avi     \n",
            "  inflating: data/fi232_xvid.avi     \n",
            "  inflating: data/fi233_xvid.avi     \n",
            "  inflating: data/fi234_xvid.avi     \n",
            "  inflating: data/fi235_xvid.avi     \n",
            "  inflating: data/fi236_xvid.avi     \n",
            "  inflating: data/fi237_xvid.avi     \n",
            "  inflating: data/fi238_xvid.avi     \n",
            "  inflating: data/fi239_xvid.avi     \n",
            "  inflating: data/fi23_xvid.avi      \n",
            "  inflating: data/fi240_xvid.avi     \n",
            "  inflating: data/fi241_xvid.avi     \n",
            "  inflating: data/fi242_xvid.avi     \n",
            "  inflating: data/fi243_xvid.avi     \n",
            "  inflating: data/fi244_xvid.avi     \n",
            "  inflating: data/fi245_xvid.avi     \n",
            "  inflating: data/fi246_xvid.avi     \n",
            "  inflating: data/fi247_xvid.avi     \n",
            "  inflating: data/fi248_xvid.avi     \n",
            "  inflating: data/fi249_xvid.avi     \n",
            "  inflating: data/fi24_xvid.avi      \n",
            "  inflating: data/fi250_xvid.avi     \n",
            "  inflating: data/fi251_xvid.avi     \n",
            "  inflating: data/fi252_xvid.avi     \n",
            "  inflating: data/fi253_xvid.avi     \n",
            "  inflating: data/fi254_xvid.avi     \n",
            "  inflating: data/fi255_xvid.avi     \n",
            "  inflating: data/fi256_xvid.avi     \n",
            "  inflating: data/fi257_xvid.avi     \n",
            "  inflating: data/fi258_xvid.avi     \n",
            "  inflating: data/fi259_xvid.avi     \n",
            "  inflating: data/fi25_xvid.avi      \n",
            "  inflating: data/fi260_xvid.avi     \n",
            "  inflating: data/fi261_xvid.avi     \n",
            "  inflating: data/fi262_xvid.avi     \n",
            "  inflating: data/fi263_xvid.avi     \n",
            "  inflating: data/fi264_xvid.avi     \n",
            "  inflating: data/fi265_xvid.avi     \n",
            "  inflating: data/fi266_xvid.avi     \n",
            "  inflating: data/fi267_xvid.avi     \n",
            "  inflating: data/fi268_xvid.avi     \n",
            "  inflating: data/fi269_xvid.avi     \n",
            "  inflating: data/fi26_xvid.avi      \n",
            "  inflating: data/fi270_xvid.avi     \n",
            "  inflating: data/fi271_xvid.avi     \n",
            "  inflating: data/fi272_xvid.avi     \n",
            "  inflating: data/fi273_xvid.avi     \n",
            "  inflating: data/fi274_xvid.avi     \n",
            "  inflating: data/fi275_xvid.avi     \n",
            "  inflating: data/fi276_xvid.avi     \n",
            "  inflating: data/fi277_xvid.avi     \n",
            "  inflating: data/fi278_xvid.avi     \n",
            "  inflating: data/fi279_xvid.avi     \n",
            "  inflating: data/fi27_xvid.avi      \n",
            "  inflating: data/fi280_xvid.avi     \n",
            "  inflating: data/fi281_xvid.avi     \n",
            "  inflating: data/fi282_xvid.avi     \n",
            "  inflating: data/fi283_xvid.avi     \n",
            "  inflating: data/fi284_xvid.avi     \n",
            "  inflating: data/fi285_xvid.avi     \n",
            "  inflating: data/fi286_xvid.avi     \n",
            "  inflating: data/fi287_xvid.avi     \n",
            "  inflating: data/fi288_xvid.avi     \n",
            "  inflating: data/fi289_xvid.avi     \n",
            "  inflating: data/fi28_xvid.avi      \n",
            "  inflating: data/fi290_xvid.avi     \n",
            "  inflating: data/fi291_xvid.avi     \n",
            "  inflating: data/fi292_xvid.avi     \n",
            "  inflating: data/fi293_xvid.avi     \n",
            "  inflating: data/fi294_xvid.avi     \n",
            "  inflating: data/fi295_xvid.avi     \n",
            "  inflating: data/fi296_xvid.avi     \n",
            "  inflating: data/fi297_xvid.avi     \n",
            "  inflating: data/fi298_xvid.avi     \n",
            "  inflating: data/fi299_xvid.avi     \n",
            "  inflating: data/fi29_xvid.avi      \n",
            "  inflating: data/fi2_xvid.avi       \n",
            "  inflating: data/fi300_xvid.avi     \n",
            "  inflating: data/fi301_xvid.avi     \n",
            "  inflating: data/fi302_xvid.avi     \n",
            "  inflating: data/fi303_xvid.avi     \n",
            "  inflating: data/fi304_xvid.avi     \n",
            "  inflating: data/fi305_xvid.avi     \n",
            "  inflating: data/fi306_xvid.avi     \n",
            "  inflating: data/fi307_xvid.avi     \n",
            "  inflating: data/fi308_xvid.avi     \n",
            "  inflating: data/fi309_xvid.avi     \n",
            "  inflating: data/fi30_xvid.avi      \n",
            "  inflating: data/fi310_xvid.avi     \n",
            "  inflating: data/fi311_xvid.avi     \n",
            "  inflating: data/fi312_xvid.avi     \n",
            "  inflating: data/fi313_xvid.avi     \n",
            "  inflating: data/fi314_xvid.avi     \n",
            "  inflating: data/fi315_xvid.avi     \n",
            "  inflating: data/fi316_xvid.avi     \n",
            "  inflating: data/fi317_xvid.avi     \n",
            "  inflating: data/fi318_xvid.avi     \n",
            "  inflating: data/fi319_xvid.avi     \n",
            "  inflating: data/fi31_xvid.avi      \n",
            "  inflating: data/fi320_xvid.avi     \n",
            "  inflating: data/fi321_xvid.avi     \n",
            "  inflating: data/fi322_xvid.avi     \n",
            "  inflating: data/fi323_xvid.avi     \n",
            "  inflating: data/fi324_xvid.avi     \n",
            "  inflating: data/fi325_xvid.avi     \n",
            "  inflating: data/fi326_xvid.avi     \n",
            "  inflating: data/fi327_xvid.avi     \n",
            "  inflating: data/fi328_xvid.avi     \n",
            "  inflating: data/fi329_xvid.avi     \n",
            "  inflating: data/fi32_xvid.avi      \n",
            "  inflating: data/fi330_xvid.avi     \n",
            "  inflating: data/fi331_xvid.avi     \n",
            "  inflating: data/fi332_xvid.avi     \n",
            "  inflating: data/fi333_xvid.avi     \n",
            "  inflating: data/fi334_xvid.avi     \n",
            "  inflating: data/fi335_xvid.avi     \n",
            "  inflating: data/fi336_xvid.avi     \n",
            "  inflating: data/fi337_xvid.avi     \n",
            "  inflating: data/fi338_xvid.avi     \n",
            "  inflating: data/fi339_xvid.avi     \n",
            "  inflating: data/fi33_xvid.avi      \n",
            "  inflating: data/fi340_xvid.avi     \n",
            "  inflating: data/fi341_xvid.avi     \n",
            "  inflating: data/fi342_xvid.avi     \n",
            "  inflating: data/fi343_xvid.avi     \n",
            "  inflating: data/fi344_xvid.avi     \n",
            "  inflating: data/fi345_xvid.avi     \n",
            "  inflating: data/fi346_xvid.avi     \n",
            "  inflating: data/fi347_xvid.avi     \n",
            "  inflating: data/fi348_xvid.avi     \n",
            "  inflating: data/fi349_xvid.avi     \n",
            "  inflating: data/fi34_xvid.avi      \n",
            "  inflating: data/fi350_xvid.avi     \n",
            "  inflating: data/fi351_xvid.avi     \n",
            "  inflating: data/fi352_xvid.avi     \n",
            "  inflating: data/fi353_xvid.avi     \n",
            "  inflating: data/fi354_xvid.avi     \n",
            "  inflating: data/fi355_xvid.avi     \n",
            "  inflating: data/fi356_xvid.avi     \n",
            "  inflating: data/fi357_xvid.avi     \n",
            "  inflating: data/fi358_xvid.avi     \n",
            "  inflating: data/fi359_xvid.avi     \n",
            "  inflating: data/fi35_xvid.avi      \n",
            "  inflating: data/fi360_xvid.avi     \n",
            "  inflating: data/fi361_xvid.avi     \n",
            "  inflating: data/fi362_xvid.avi     \n",
            "  inflating: data/fi363_xvid.avi     \n",
            "  inflating: data/fi364_xvid.avi     \n",
            "  inflating: data/fi365_xvid.avi     \n",
            "  inflating: data/fi366_xvid.avi     \n",
            "  inflating: data/fi367_xvid.avi     \n",
            "  inflating: data/fi368_xvid.avi     \n",
            "  inflating: data/fi369_xvid.avi     \n",
            "  inflating: data/fi36_xvid.avi      \n",
            "  inflating: data/fi370_xvid.avi     \n",
            "  inflating: data/fi371_xvid.avi     \n",
            "  inflating: data/fi372_xvid.avi     \n",
            "  inflating: data/fi373_xvid.avi     \n",
            "  inflating: data/fi374_xvid.avi     \n",
            "  inflating: data/fi375_xvid.avi     \n",
            "  inflating: data/fi376_xvid.avi     \n",
            "  inflating: data/fi377_xvid.avi     \n",
            "  inflating: data/fi378_xvid.avi     \n",
            "  inflating: data/fi379_xvid.avi     \n",
            "  inflating: data/fi37_xvid.avi      \n",
            "  inflating: data/fi380_xvid.avi     \n",
            "  inflating: data/fi381_xvid.avi     \n",
            "  inflating: data/fi382_xvid.avi     \n",
            "  inflating: data/fi383_xvid.avi     \n",
            "  inflating: data/fi384_xvid.avi     \n",
            "  inflating: data/fi385_xvid.avi     \n",
            "  inflating: data/fi386_xvid.avi     \n",
            "  inflating: data/fi387_xvid.avi     \n",
            "  inflating: data/fi388_xvid.avi     \n",
            "  inflating: data/fi389_xvid.avi     \n",
            "  inflating: data/fi38_xvid.avi      \n",
            "  inflating: data/fi390_xvid.avi     \n",
            "  inflating: data/fi391_xvid.avi     \n",
            "  inflating: data/fi392_xvid.avi     \n",
            "  inflating: data/fi393_xvid.avi     \n",
            "  inflating: data/fi394_xvid.avi     \n",
            "  inflating: data/fi395_xvid.avi     \n",
            "  inflating: data/fi396_xvid.avi     \n",
            "  inflating: data/fi397_xvid.avi     \n",
            "  inflating: data/fi398_xvid.avi     \n",
            "  inflating: data/fi399_xvid.avi     \n",
            "  inflating: data/fi39_xvid.avi      \n",
            "  inflating: data/fi3_xvid.avi       \n",
            "  inflating: data/fi400_xvid.avi     \n",
            "  inflating: data/fi401_xvid.avi     \n",
            "  inflating: data/fi402_xvid.avi     \n",
            "  inflating: data/fi403_xvid.avi     \n",
            "  inflating: data/fi404_xvid.avi     \n",
            "  inflating: data/fi405_xvid.avi     \n",
            "  inflating: data/fi406_xvid.avi     \n",
            "  inflating: data/fi407_xvid.avi     \n",
            "  inflating: data/fi408_xvid.avi     \n",
            "  inflating: data/fi409_xvid.avi     \n",
            "  inflating: data/fi40_xvid.avi      \n",
            "  inflating: data/fi410_xvid.avi     \n",
            "  inflating: data/fi411_xvid.avi     \n",
            "  inflating: data/fi412_xvid.avi     \n",
            "  inflating: data/fi413_xvid.avi     \n",
            "  inflating: data/fi414_xvid.avi     \n",
            "  inflating: data/fi415_xvid.avi     \n",
            "  inflating: data/fi416_xvid.avi     \n",
            "  inflating: data/fi417_xvid.avi     \n",
            "  inflating: data/fi418_xvid.avi     \n",
            "  inflating: data/fi419_xvid.avi     \n",
            "  inflating: data/fi41_xvid.avi      \n",
            "  inflating: data/fi420_xvid.avi     \n",
            "  inflating: data/fi421_xvid.avi     \n",
            "  inflating: data/fi422_xvid.avi     \n",
            "  inflating: data/fi423_xvid.avi     \n",
            "  inflating: data/fi424_xvid.avi     \n",
            "  inflating: data/fi425_xvid.avi     \n",
            "  inflating: data/fi426_xvid.avi     \n",
            "  inflating: data/fi427_xvid.avi     \n",
            "  inflating: data/fi428_xvid.avi     \n",
            "  inflating: data/fi429_xvid.avi     \n",
            "  inflating: data/fi42_xvid.avi      \n",
            "  inflating: data/fi430_xvid.avi     \n",
            "  inflating: data/fi431_xvid.avi     \n",
            "  inflating: data/fi432_xvid.avi     \n",
            "  inflating: data/fi433_xvid.avi     \n",
            "  inflating: data/fi434_xvid.avi     \n",
            "  inflating: data/fi435_xvid.avi     \n",
            "  inflating: data/fi436_xvid.avi     \n",
            "  inflating: data/fi437_xvid.avi     \n",
            "  inflating: data/fi438_xvid.avi     \n",
            "  inflating: data/fi439_xvid.avi     \n",
            "  inflating: data/fi43_xvid.avi      \n",
            "  inflating: data/fi440_xvid.avi     \n",
            "  inflating: data/fi441_xvid.avi     \n",
            "  inflating: data/fi442_xvid.avi     \n",
            "  inflating: data/fi443_xvid.avi     \n",
            "  inflating: data/fi444_xvid.avi     \n",
            "  inflating: data/fi445_xvid.avi     \n",
            "  inflating: data/fi446_xvid.avi     \n",
            "  inflating: data/fi447_xvid.avi     \n",
            "  inflating: data/fi448_xvid.avi     \n",
            "  inflating: data/fi449_xvid.avi     \n",
            "  inflating: data/fi44_xvid.avi      \n",
            "  inflating: data/fi450_xvid.avi     \n",
            "  inflating: data/fi451_xvid.avi     \n",
            "  inflating: data/fi452_xvid.avi     \n",
            "  inflating: data/fi453_xvid.avi     \n",
            "  inflating: data/fi454_xvid.avi     \n",
            "  inflating: data/fi455_xvid.avi     \n",
            "  inflating: data/fi456_xvid.avi     \n",
            "  inflating: data/fi457_xvid.avi     \n",
            "  inflating: data/fi458_xvid.avi     \n",
            "  inflating: data/fi459_xvid.avi     \n",
            "  inflating: data/fi45_xvid.avi      \n",
            "  inflating: data/fi460_xvid.avi     \n",
            "  inflating: data/fi461_xvid.avi     \n",
            "  inflating: data/fi462_xvid.avi     \n",
            "  inflating: data/fi463_xvid.avi     \n",
            "  inflating: data/fi464_xvid.avi     \n",
            "  inflating: data/fi465_xvid.avi     \n",
            "  inflating: data/fi466_xvid.avi     \n",
            "  inflating: data/fi467_xvid.avi     \n",
            "  inflating: data/fi468_xvid.avi     \n",
            "  inflating: data/fi469_xvid.avi     \n",
            "  inflating: data/fi46_xvid.avi      \n",
            "  inflating: data/fi470_xvid.avi     \n",
            "  inflating: data/fi471_xvid.avi     \n",
            "  inflating: data/fi472_xvid.avi     \n",
            "  inflating: data/fi473_xvid.avi     \n",
            "  inflating: data/fi474_xvid.avi     \n",
            "  inflating: data/fi475_xvid.avi     \n",
            "  inflating: data/fi476_xvid.avi     \n",
            "  inflating: data/fi477_xvid.avi     \n",
            "  inflating: data/fi478_xvid.avi     \n",
            "  inflating: data/fi479_xvid.avi     \n",
            "  inflating: data/fi47_xvid.avi      \n",
            "  inflating: data/fi480_xvid.avi     \n",
            "  inflating: data/fi481_xvid.avi     \n",
            "  inflating: data/fi482_xvid.avi     \n",
            "  inflating: data/fi483_xvid.avi     \n",
            "  inflating: data/fi484_xvid.avi     \n",
            "  inflating: data/fi485_xvid.avi     \n",
            "  inflating: data/fi486_xvid.avi     \n",
            "  inflating: data/fi487_xvid.avi     \n",
            "  inflating: data/fi488_xvid.avi     \n",
            "  inflating: data/fi489_xvid.avi     \n",
            "  inflating: data/fi48_xvid.avi      \n",
            "  inflating: data/fi490_xvid.avi     \n",
            "  inflating: data/fi491_xvid.avi     \n",
            "  inflating: data/fi492_xvid.avi     \n",
            "  inflating: data/fi493_xvid.avi     \n",
            "  inflating: data/fi494_xvid.avi     \n",
            "  inflating: data/fi495_xvid.avi     \n",
            "  inflating: data/fi496_xvid.avi     \n",
            "  inflating: data/fi497_xvid.avi     \n",
            "  inflating: data/fi498_xvid.avi     \n",
            "  inflating: data/fi499_xvid.avi     \n",
            "  inflating: data/fi49_xvid.avi      \n",
            "  inflating: data/fi4_xvid.avi       \n",
            "  inflating: data/fi500_xvid.avi     \n",
            "  inflating: data/fi50_xvid.avi      \n",
            "  inflating: data/fi51_xvid.avi      \n",
            "  inflating: data/fi52_xvid.avi      \n",
            "  inflating: data/fi53_xvid.avi      \n",
            "  inflating: data/fi54_xvid.avi      \n",
            "  inflating: data/fi55_xvid.avi      \n",
            "  inflating: data/fi56_xvid.avi      \n",
            "  inflating: data/fi57_xvid.avi      \n",
            "  inflating: data/fi58_xvid.avi      \n",
            "  inflating: data/fi59_xvid.avi      \n",
            "  inflating: data/fi5_xvid.avi       \n",
            "  inflating: data/fi60_xvid.avi      \n",
            "  inflating: data/fi61_xvid.avi      \n",
            "  inflating: data/fi62_xvid.avi      \n",
            "  inflating: data/fi63_xvid.avi      \n",
            "  inflating: data/fi64_xvid.avi      \n",
            "  inflating: data/fi65_xvid.avi      \n",
            "  inflating: data/fi66_xvid.avi      \n",
            "  inflating: data/fi67_xvid.avi      \n",
            "  inflating: data/fi68_xvid.avi      \n",
            "  inflating: data/fi69_xvid.avi      \n",
            "  inflating: data/fi6_xvid.avi       \n",
            "  inflating: data/fi70_xvid.avi      \n",
            "  inflating: data/fi71_xvid.avi      \n",
            "  inflating: data/fi72_xvid.avi      \n",
            "  inflating: data/fi73_xvid.avi      \n",
            "  inflating: data/fi74_xvid.avi      \n",
            "  inflating: data/fi75_xvid.avi      \n",
            "  inflating: data/fi76_xvid.avi      \n",
            "  inflating: data/fi77_xvid.avi      \n",
            "  inflating: data/fi78_xvid.avi      \n",
            "  inflating: data/fi79_xvid.avi      \n",
            "  inflating: data/fi7_xvid.avi       \n",
            "  inflating: data/fi80_xvid.avi      \n",
            "  inflating: data/fi81_xvid.avi      \n",
            "  inflating: data/fi82_xvid.avi      \n",
            "  inflating: data/fi83_xvid.avi      \n",
            "  inflating: data/fi84_xvid.avi      \n",
            "  inflating: data/fi85_xvid.avi      \n",
            "  inflating: data/fi86_xvid.avi      \n",
            "  inflating: data/fi87_xvid.avi      \n",
            "  inflating: data/fi88_xvid.avi      \n",
            "  inflating: data/fi89_xvid.avi      \n",
            "  inflating: data/fi8_xvid.avi       \n",
            "  inflating: data/fi90_xvid.avi      \n",
            "  inflating: data/fi91_xvid.avi      \n",
            "  inflating: data/fi92_xvid.avi      \n",
            "  inflating: data/fi93_xvid.avi      \n",
            "  inflating: data/fi94_xvid.avi      \n",
            "  inflating: data/fi95_xvid.avi      \n",
            "  inflating: data/fi96_xvid.avi      \n",
            "  inflating: data/fi97_xvid.avi      \n",
            "  inflating: data/fi98_xvid.avi      \n",
            "  inflating: data/fi99_xvid.avi      \n",
            "  inflating: data/fi9_xvid.avi       \n",
            "  inflating: data/no100_xvid.avi     \n",
            "  inflating: data/no101_xvid.avi     \n",
            "  inflating: data/no102_xvid.avi     \n",
            "  inflating: data/no103_xvid.avi     \n",
            "  inflating: data/no104_xvid.avi     \n",
            "  inflating: data/no105_xvid.avi     \n",
            "  inflating: data/no106_xvid.avi     \n",
            "  inflating: data/no107_xvid.avi     \n",
            "  inflating: data/no108_xvid.avi     \n",
            "  inflating: data/no109_xvid.avi     \n",
            "  inflating: data/no10_xvid.avi      \n",
            "  inflating: data/no110_xvid.avi     \n",
            "  inflating: data/no111_xvid.avi     \n",
            "  inflating: data/no112_xvid.avi     \n",
            "  inflating: data/no113_xvid.avi     \n",
            "  inflating: data/no114_xvid.avi     \n",
            "  inflating: data/no115_xvid.avi     \n",
            "  inflating: data/no116_xvid.avi     \n",
            "  inflating: data/no117_xvid.avi     \n",
            "  inflating: data/no118_xvid.avi     \n",
            "  inflating: data/no119_xvid.avi     \n",
            "  inflating: data/no11_xvid.avi      \n",
            "  inflating: data/no120_xvid.avi     \n",
            "  inflating: data/no121_xvid.avi     \n",
            "  inflating: data/no122_xvid.avi     \n",
            "  inflating: data/no123_xvid.avi     \n",
            "  inflating: data/no124_xvid.avi     \n",
            "  inflating: data/no125_xvid.avi     \n",
            "  inflating: data/no126_xvid.avi     \n",
            "  inflating: data/no127_xvid.avi     \n",
            "  inflating: data/no128_xvid.avi     \n",
            "  inflating: data/no129_xvid.avi     \n",
            "  inflating: data/no12_xvid.avi      \n",
            "  inflating: data/no130_xvid.avi     \n",
            "  inflating: data/no131_xvid.avi     \n",
            "  inflating: data/no132_xvid.avi     \n",
            "  inflating: data/no133_xvid.avi     \n",
            "  inflating: data/no134_xvid.avi     \n",
            "  inflating: data/no135_xvid.avi     \n",
            "  inflating: data/no136_xvid.avi     \n",
            "  inflating: data/no137_xvid.avi     \n",
            "  inflating: data/no138_xvid.avi     \n",
            "  inflating: data/no139_xvid.avi     \n",
            "  inflating: data/no13_xvid.avi      \n",
            "  inflating: data/no140_xvid.avi     \n",
            "  inflating: data/no141_xvid.avi     \n",
            "  inflating: data/no142__xvid.avi    \n",
            "  inflating: data/no143_xvid.avi     \n",
            "  inflating: data/no144_xvid.avi     \n",
            "  inflating: data/no145_xvid.avi     \n",
            "  inflating: data/no146_xvid.avi     \n",
            "  inflating: data/no147_xvid.avi     \n",
            "  inflating: data/no148_xvid.avi     \n",
            "  inflating: data/no149_xvid.avi     \n",
            "  inflating: data/no14_xvid.avi      \n",
            "  inflating: data/no150_xvid.avi     \n",
            "  inflating: data/no151_xvid.avi     \n",
            "  inflating: data/no152_xvid.avi     \n",
            "  inflating: data/no153_xvid.avi     \n",
            "  inflating: data/no154_xvid.avi     \n",
            "  inflating: data/no155_xvid.avi     \n",
            "  inflating: data/no156_xvid.avi     \n",
            "  inflating: data/no157_xvid.avi     \n",
            "  inflating: data/no158_xvid.avi     \n",
            "  inflating: data/no159_xvid.avi     \n",
            "  inflating: data/no15_xvid.avi      \n",
            "  inflating: data/no160_xvid.avi     \n",
            "  inflating: data/no161_xvid.avi     \n",
            "  inflating: data/no162_xvid.avi     \n",
            "  inflating: data/no163_xvid.avi     \n",
            "  inflating: data/no164_xvid.avi     \n",
            "  inflating: data/no165_xvid.avi     \n",
            "  inflating: data/no166_xvid.avi     \n",
            "  inflating: data/no167_xvid.avi     \n",
            "  inflating: data/no168_xvid.avi     \n",
            "  inflating: data/no169_xvid.avi     \n",
            "  inflating: data/no16_xvid.avi      \n",
            "  inflating: data/no170_xvid.avi     \n",
            "  inflating: data/no171_xvid.avi     \n",
            "  inflating: data/no172_xvid.avi     \n",
            "  inflating: data/no173_xvid.avi     \n",
            "  inflating: data/no174_xvid.avi     \n",
            "  inflating: data/no175_xvid.avi     \n",
            "  inflating: data/no176_xvid.avi     \n",
            "  inflating: data/no177_xvid.avi     \n",
            "  inflating: data/no178_xvid.avi     \n",
            "  inflating: data/no179_xvid.avi     \n",
            "  inflating: data/no17_xvid.avi      \n",
            "  inflating: data/no180_xvid.avi     \n",
            "  inflating: data/no181_xvid.avi     \n",
            "  inflating: data/no182_xvid.avi     \n",
            "  inflating: data/no183_xvid.avi     \n",
            "  inflating: data/no184_xvid.avi     \n",
            "  inflating: data/no185_xvid.avi     \n",
            "  inflating: data/no186_xvid.avi     \n",
            "  inflating: data/no187_xvid.avi     \n",
            "  inflating: data/no188_xvid.avi     \n",
            "  inflating: data/no189_xvid.avi     \n",
            "  inflating: data/no18_xvid.avi      \n",
            "  inflating: data/no190_xvid.avi     \n",
            "  inflating: data/no191_xvid.avi     \n",
            "  inflating: data/no192_xvid.avi     \n",
            "  inflating: data/no193_xvid.avi     \n",
            "  inflating: data/no194_xvid.avi     \n",
            "  inflating: data/no195_xvid.avi     \n",
            "  inflating: data/no196_xvid.avi     \n",
            "  inflating: data/no197_xvid.avi     \n",
            "  inflating: data/no198_xvid.avi     \n",
            "  inflating: data/no199_xvid.avi     \n",
            "  inflating: data/no19_xvid.avi      \n",
            "  inflating: data/no1_xvid.avi       \n",
            "  inflating: data/no200_xvid.avi     \n",
            "  inflating: data/no201_xvid.avi     \n",
            "  inflating: data/no202_xvid.avi     \n",
            "  inflating: data/no203_xvid.avi     \n",
            "  inflating: data/no204_xvid.avi     \n",
            "  inflating: data/no205_xvid.avi     \n",
            "  inflating: data/no206_xvid.avi     \n",
            "  inflating: data/no207_xvid.avi     \n",
            "  inflating: data/no208_xvid.avi     \n",
            "  inflating: data/no209_xvid.avi     \n",
            "  inflating: data/no20_xvid.avi      \n",
            "  inflating: data/no210_xvid.avi     \n",
            "  inflating: data/no211_xvid.avi     \n",
            "  inflating: data/no212_xvid.avi     \n",
            "  inflating: data/no213_xvid.avi     \n",
            "  inflating: data/no214_xvid.avi     \n",
            "  inflating: data/no215_xvid.avi     \n",
            "  inflating: data/no216_xvid.avi     \n",
            "  inflating: data/no217_xvid.avi     \n",
            "  inflating: data/no218_xvid.avi     \n",
            "  inflating: data/no219_xvid.avi     \n",
            "  inflating: data/no21_xvid.avi      \n",
            "  inflating: data/no220_xvid.avi     \n",
            "  inflating: data/no221_xvid.avi     \n",
            "  inflating: data/no222_xvid.avi     \n",
            "  inflating: data/no223_xvid.avi     \n",
            "  inflating: data/no224_xvid.avi     \n",
            "  inflating: data/no225_xvid.avi     \n",
            "  inflating: data/no226_xvid.avi     \n",
            "  inflating: data/no227_xvid.avi     \n",
            "  inflating: data/no228_xvid.avi     \n",
            "  inflating: data/no229_xvid.avi     \n",
            "  inflating: data/no22_xvid.avi      \n",
            "  inflating: data/no230_xvid.avi     \n",
            "  inflating: data/no231_xvid.avi     \n",
            "  inflating: data/no232_xvid.avi     \n",
            "  inflating: data/no233_xvid.avi     \n",
            "  inflating: data/no234_xvid.avi     \n",
            "  inflating: data/no235_xvid.avi     \n",
            "  inflating: data/no236_xvid.avi     \n",
            "  inflating: data/no237_xvid.avi     \n",
            "  inflating: data/no238mpg_xvid.avi  \n",
            "  inflating: data/no239_xvid.avi     \n",
            "  inflating: data/no23_xvid.avi      \n",
            "  inflating: data/no240_xvid.avi     \n",
            "  inflating: data/no241_xvid.avi     \n",
            "  inflating: data/no242_xvid.avi     \n",
            "  inflating: data/no243_xvid.avi     \n",
            "  inflating: data/no244_xvid.avi     \n",
            "  inflating: data/no245_xvid.avi     \n",
            "  inflating: data/no246_xvid.avi     \n",
            "  inflating: data/no247_xvid.avi     \n",
            "  inflating: data/no248_xvid.avi     \n",
            "  inflating: data/no249_xvid.avi     \n",
            "  inflating: data/no24_xvid.avi      \n",
            "  inflating: data/no250_xvid.avi     \n",
            "  inflating: data/no251_xvid.avi     \n",
            "  inflating: data/no252_xvid.avi     \n",
            "  inflating: data/no253_xvid.avi     \n",
            "  inflating: data/no254_xvid.avi     \n",
            "  inflating: data/no255_xvid.avi     \n",
            "  inflating: data/no256_xvid.avi     \n",
            "  inflating: data/no257_xvid.avi     \n",
            "  inflating: data/no258_xvid.avi     \n",
            "  inflating: data/no259_xvid.avi     \n",
            "  inflating: data/no25_xvid.avi      \n",
            "  inflating: data/no260_xvid.avi     \n",
            "  inflating: data/no261_xvid.avi     \n",
            "  inflating: data/no262_xvid.avi     \n",
            "  inflating: data/no263_xvid.avi     \n",
            "  inflating: data/no264_xvid.avi     \n",
            "  inflating: data/no265_xvid.avi     \n",
            "  inflating: data/no266_xvid.avi     \n",
            "  inflating: data/no267_xvid.avi     \n",
            "  inflating: data/no268_xvid.avi     \n",
            "  inflating: data/no269_xvid.avi     \n",
            "  inflating: data/no26_xvid.avi      \n",
            "  inflating: data/no270_xvid.avi     \n",
            "  inflating: data/no271_xvid.avi     \n",
            "  inflating: data/no272_xvid.avi     \n",
            "  inflating: data/no273_xvid.avi     \n",
            "  inflating: data/no274_xvid.avi     \n",
            "  inflating: data/no275_xvid.avi     \n",
            "  inflating: data/no276_xvid.avi     \n",
            "  inflating: data/no277_xvid.avi     \n",
            "  inflating: data/no278_xvid.avi     \n",
            "  inflating: data/no279_xvid.avi     \n",
            "  inflating: data/no27_xvid.avi      \n",
            "  inflating: data/no280_xvid.avi     \n",
            "  inflating: data/no281_xvid.avi     \n",
            "  inflating: data/no282_xvid.avi     \n",
            "  inflating: data/no283_xvid.avi     \n",
            "  inflating: data/no284_xvid.avi     \n",
            "  inflating: data/no285_xvid.avi     \n",
            "  inflating: data/no286_xvid.avi     \n",
            "  inflating: data/no287_xvid.avi     \n",
            "  inflating: data/no288_xvid.avi     \n",
            "  inflating: data/no289_xvid.avi     \n",
            "  inflating: data/no28_xvid.avi      \n",
            "  inflating: data/no290_xvid.avi     \n",
            "  inflating: data/no291_xvid.avi     \n",
            "  inflating: data/no292_xvid.avi     \n",
            "  inflating: data/no293_xvid.avi     \n",
            "  inflating: data/no294_xvid.avi     \n",
            "  inflating: data/no295_xvid.avi     \n",
            "  inflating: data/no296_xvid.avi     \n",
            "  inflating: data/no297_xvid.avi     \n",
            "  inflating: data/no298_xvid.avi     \n",
            "  inflating: data/no299_xvid.avi     \n",
            "  inflating: data/no29_xvid.avi      \n",
            "  inflating: data/no2_xvid.avi       \n",
            "  inflating: data/no300_xvid.avi     \n",
            "  inflating: data/no301_xvid.avi     \n",
            "  inflating: data/no302_xvid.avi     \n",
            "  inflating: data/no303_xvid.avi     \n",
            "  inflating: data/no304_xvid.avi     \n",
            "  inflating: data/no305_xvid.avi     \n",
            "  inflating: data/no306_xvid.avi     \n",
            "  inflating: data/no307_xvid.avi     \n",
            "  inflating: data/no308_xvid.avi     \n",
            "  inflating: data/no309_xvid.avi     \n",
            "  inflating: data/no30_xvid.avi      \n",
            "  inflating: data/no310_xvid.avi     \n",
            "  inflating: data/no311_xvid.avi     \n",
            "  inflating: data/no312_xvid.avi     \n",
            "  inflating: data/no313_xvid.avi     \n",
            "  inflating: data/no314_xvid.avi     \n",
            "  inflating: data/no315_xvid.avi     \n",
            "  inflating: data/no316_xvid.avi     \n",
            "  inflating: data/no317_xvid.avi     \n",
            "  inflating: data/no318_xvid.avi     \n",
            "  inflating: data/no319_xvid.avi     \n",
            "  inflating: data/no31_xvid.avi      \n",
            "  inflating: data/no320_xvid.avi     \n",
            "  inflating: data/no321_xvid.avi     \n",
            "  inflating: data/no322_xvid.avi     \n",
            "  inflating: data/no323_xvid.avi     \n",
            "  inflating: data/no324_xvid.avi     \n",
            "  inflating: data/no325_xvid.avi     \n",
            "  inflating: data/no326_xvid.avi     \n",
            "  inflating: data/no327_xvid.avi     \n",
            "  inflating: data/no328_xvid.avi     \n",
            "  inflating: data/no329_xvid.avi     \n",
            "  inflating: data/no32_xvid.avi      \n",
            "  inflating: data/no330_xvid.avi     \n",
            "  inflating: data/no331_xvid.avi     \n",
            "  inflating: data/no332_xvid.avi     \n",
            "  inflating: data/no333_xvid.avi     \n",
            "  inflating: data/no334_xvid.avi     \n",
            "  inflating: data/no335_xvid.avi     \n",
            "  inflating: data/no336_xvid.avi     \n",
            "  inflating: data/no337_xvid.avi     \n",
            "  inflating: data/no338_xvid.avi     \n",
            "  inflating: data/no339_xvid.avi     \n",
            "  inflating: data/no33_xvid.avi      \n",
            "  inflating: data/no340_xvid.avi     \n",
            "  inflating: data/no341_xvid.avi     \n",
            "  inflating: data/no342_xvid.avi     \n",
            "  inflating: data/no343_xvid.avi     \n",
            "  inflating: data/no344_xvid.avi     \n",
            "  inflating: data/no345_xvid.avi     \n",
            "  inflating: data/no346_xvid.avi     \n",
            "  inflating: data/no347_xvid.avi     \n",
            "  inflating: data/no348_xvid.avi     \n",
            "  inflating: data/no349_xvid.avi     \n",
            "  inflating: data/no34_xvid.avi      \n",
            "  inflating: data/no350_xvid.avi     \n",
            "  inflating: data/no351_xvid.avi     \n",
            "  inflating: data/no352_xvid.avi     \n",
            "  inflating: data/no353_xvid.avi     \n",
            "  inflating: data/no354_xvid.avi     \n",
            "  inflating: data/no355_xvid.avi     \n",
            "  inflating: data/no356_xvid.avi     \n",
            "  inflating: data/no357_xvid.avi     \n",
            "  inflating: data/no358_xvid.avi     \n",
            "  inflating: data/no359_xvid.avi     \n",
            "  inflating: data/no35_xvid.avi      \n",
            "  inflating: data/no360_xvid.avi     \n",
            "  inflating: data/no361_xvid.avi     \n",
            "  inflating: data/no362_xvid.avi     \n",
            "  inflating: data/no363_xvid.avi     \n",
            "  inflating: data/no364_xvid.avi     \n",
            "  inflating: data/no365_xvid.avi     \n",
            "  inflating: data/no366_xvid.avi     \n",
            "  inflating: data/no367_xvid.avi     \n",
            "  inflating: data/no368_xvid.avi     \n",
            "  inflating: data/no369_xvid.avi     \n",
            "  inflating: data/no36_xvid.avi      \n",
            "  inflating: data/no370_xvid.avi     \n",
            "  inflating: data/no371_xvid.avi     \n",
            "  inflating: data/no372_xvid.avi     \n",
            "  inflating: data/no373_xvid.avi     \n",
            "  inflating: data/no374_xvid.avi     \n",
            "  inflating: data/no375_xvid.avi     \n",
            "  inflating: data/no376_xvid.avi     \n",
            "  inflating: data/no377_xvid.avi     \n",
            "  inflating: data/no378_xvid.avi     \n",
            "  inflating: data/no379_xvid.avi     \n",
            "  inflating: data/no37_xvid.avi      \n",
            "  inflating: data/no380_xvid.avi     \n",
            "  inflating: data/no381_xvid.avi     \n",
            "  inflating: data/no382_xvid.avi     \n",
            "  inflating: data/no383_xvid.avi     \n",
            "  inflating: data/no384_xvid.avi     \n",
            "  inflating: data/no385_xvid.avi     \n",
            "  inflating: data/no386_xvid.avi     \n",
            "  inflating: data/no387_xvid.avi     \n",
            "  inflating: data/no388_xvid.avi     \n",
            "  inflating: data/no389_xvid.avi     \n",
            "  inflating: data/no38_xvid.avi      \n",
            "  inflating: data/no390_xvid.avi     \n",
            "  inflating: data/no391_xvid.avi     \n",
            "  inflating: data/no392_xvid.avi     \n",
            "  inflating: data/no393_xvid.avi     \n",
            "  inflating: data/no394_xvid.avi     \n",
            "  inflating: data/no395_xvid.avi     \n",
            "  inflating: data/no396_xvid.avi     \n",
            "  inflating: data/no397_xvid.avi     \n",
            "  inflating: data/no398_xvid.avi     \n",
            "  inflating: data/no399_xvid.avi     \n",
            "  inflating: data/no39_xvid.avi      \n",
            "  inflating: data/no3_xvid.avi       \n",
            "  inflating: data/no400_xvid.avi     \n",
            "  inflating: data/no401_xvid.avi     \n",
            "  inflating: data/no402_xvid.avi     \n",
            "  inflating: data/no403_xvid.avi     \n",
            "  inflating: data/no404_xvid.avi     \n",
            "  inflating: data/no405_xvid.avi     \n",
            "  inflating: data/no406_xvid.avi     \n",
            "  inflating: data/no407_xvid.avi     \n",
            "  inflating: data/no408_xvid.avi     \n",
            "  inflating: data/no409_xvid.avi     \n",
            "  inflating: data/no40_xvid.avi      \n",
            "  inflating: data/no410_xvid.avi     \n",
            "  inflating: data/no411_xvid.avi     \n",
            "  inflating: data/no412_xvid.avi     \n",
            "  inflating: data/no413_xvid.avi     \n",
            "  inflating: data/no414_xvid.avi     \n",
            "  inflating: data/no415_xvid.avi     \n",
            "  inflating: data/no416_xvid.avi     \n",
            "  inflating: data/no417_xvid.avi     \n",
            "  inflating: data/no418_xvid.avi     \n",
            "  inflating: data/no419_xvid.avi     \n",
            "  inflating: data/no41_xvid.avi      \n",
            "  inflating: data/no420_xvid.avi     \n",
            "  inflating: data/no421_xvid.avi     \n",
            "  inflating: data/no422_xvid.avi     \n",
            "  inflating: data/no423_xvid.avi     \n",
            "  inflating: data/no424_xvid.avi     \n",
            "  inflating: data/no425_xvid.avi     \n",
            "  inflating: data/no426_xvid.avi     \n",
            "  inflating: data/no427_xvid.avi     \n",
            "  inflating: data/no428_xvid.avi     \n",
            "  inflating: data/no429_xvid.avi     \n",
            "  inflating: data/no42_xvid.avi      \n",
            "  inflating: data/no430_xvid.avi     \n",
            "  inflating: data/no431_xvid.avi     \n",
            "  inflating: data/no432_xvid.avi     \n",
            "  inflating: data/no433_xvid.avi     \n",
            "  inflating: data/no434_xvid.avi     \n",
            "  inflating: data/no435_xvid.avi     \n",
            "  inflating: data/no436_xvid.avi     \n",
            "  inflating: data/no437_xvid.avi     \n",
            "  inflating: data/no438_xvid.avi     \n",
            "  inflating: data/no439_xvid.avi     \n",
            "  inflating: data/no43_xvid.avi      \n",
            "  inflating: data/no440_xvid.avi     \n",
            "  inflating: data/no441_xvid.avi     \n",
            "  inflating: data/no442_xvid.avi     \n",
            "  inflating: data/no443_xvid.avi     \n",
            "  inflating: data/no444_xvid.avi     \n",
            "  inflating: data/no445_xvid.avi     \n",
            "  inflating: data/no446_xvid.avi     \n",
            "  inflating: data/no447_xvid.avi     \n",
            "  inflating: data/no448_xvid.avi     \n",
            "  inflating: data/no449_xvid.avi     \n",
            "  inflating: data/no44_xvid.avi      \n",
            "  inflating: data/no450_xvid.avi     \n",
            "  inflating: data/no451_xvid.avi     \n",
            "  inflating: data/no452_xvid.avi     \n",
            "  inflating: data/no453_xvid.avi     \n",
            "  inflating: data/no454_xvid.avi     \n",
            "  inflating: data/no455_xvid.avi     \n",
            "  inflating: data/no456_xvid.avi     \n",
            "  inflating: data/no457_xvid.avi     \n",
            "  inflating: data/no458_xvid.avi     \n",
            "  inflating: data/no459_xvid.avi     \n",
            "  inflating: data/no45_xvid.avi      \n",
            "  inflating: data/no460_xvid.avi     \n",
            "  inflating: data/no461_xvid.avi     \n",
            "  inflating: data/no462_xvid.avi     \n",
            "  inflating: data/no463_xvid.avi     \n",
            "  inflating: data/no464_xvid.avi     \n",
            "  inflating: data/no465_xvid.avi     \n",
            "  inflating: data/no466_xvid.avi     \n",
            "  inflating: data/no467_xvid.avi     \n",
            "  inflating: data/no468_xvid.avi     \n",
            "  inflating: data/no469_xvid.avi     \n",
            "  inflating: data/no46_xvid.avi      \n",
            "  inflating: data/no470_xvid.avi     \n",
            "  inflating: data/no471_xvid.avi     \n",
            "  inflating: data/no472_xvid.avi     \n",
            "  inflating: data/no473_xvid.avi     \n",
            "  inflating: data/no474_xvid.avi     \n",
            "  inflating: data/no475_xvid.avi     \n",
            "  inflating: data/no476_xvid.avi     \n",
            "  inflating: data/no477_xvid.avi     \n",
            "  inflating: data/no478_xvid.avi     \n",
            "  inflating: data/no479_xvid.avi     \n",
            "  inflating: data/no47_xvid.avi      \n",
            "  inflating: data/no480_xvid.avi     \n",
            "  inflating: data/no481_xvid.avi     \n",
            "  inflating: data/no482_xvid.avi     \n",
            "  inflating: data/no483_xvid.avi     \n",
            "  inflating: data/no484_xvid.avi     \n",
            "  inflating: data/no485_xvid.avi     \n",
            "  inflating: data/no486_xvid.avi     \n",
            "  inflating: data/no487_xvid.avi     \n",
            "  inflating: data/no488_xvid.avi     \n",
            "  inflating: data/no489_xvid.avi     \n",
            "  inflating: data/no48_xvid.avi      \n",
            "  inflating: data/no490_xvid.avi     \n",
            "  inflating: data/no491_xvid.avi     \n",
            "  inflating: data/no492_xvid.avi     \n",
            "  inflating: data/no493_xvid.avi     \n",
            "  inflating: data/no494_xvid.avi     \n",
            "  inflating: data/no495_xvid.avi     \n",
            "  inflating: data/no496_xvid.avi     \n",
            "  inflating: data/no497_xvid.avi     \n",
            "  inflating: data/no498_xvid.avi     \n",
            "  inflating: data/no499_xvid.avi     \n",
            "  inflating: data/no49_xvid.avi      \n",
            "  inflating: data/no4_xvid.avi       \n",
            "  inflating: data/no500_xvid.avi     \n",
            "  inflating: data/no50_xvid.avi      \n",
            "  inflating: data/no51_xvid.avi      \n",
            "  inflating: data/no52_xvid.avi      \n",
            "  inflating: data/no53_xvid.avi      \n",
            "  inflating: data/no54_xvid.avi      \n",
            "  inflating: data/no55_xvid.avi      \n",
            "  inflating: data/no56_xvid.avi      \n",
            "  inflating: data/no57_xvid.avi      \n",
            "  inflating: data/no58_xvid.avi      \n",
            "  inflating: data/no59_xvid.avi      \n",
            "  inflating: data/no5_xvid.avi       \n",
            "  inflating: data/no60_xvid.avi      \n",
            "  inflating: data/no61_xvid.avi      \n",
            "  inflating: data/no62_xvid.avi      \n",
            "  inflating: data/no63_xvid.avi      \n",
            "  inflating: data/no64_xvid.avi      \n",
            "  inflating: data/no65___xvid.avi    \n",
            "  inflating: data/no66_xvid.avi      \n",
            "  inflating: data/no67_xvid.avi      \n",
            "  inflating: data/no68_xvid.avi      \n",
            "  inflating: data/no69_xvid.avi      \n",
            "  inflating: data/no6_xvid.avi       \n",
            "  inflating: data/no70_xvid.avi      \n",
            "  inflating: data/no71_xvid.avi      \n",
            "  inflating: data/no72_xvid.avi      \n",
            "  inflating: data/no73_xvid.avi      \n",
            "  inflating: data/no74_xvid.avi      \n",
            "  inflating: data/no75_xvid.avi      \n",
            "  inflating: data/no76_xvid.avi      \n",
            "  inflating: data/no77_xvid.avi      \n",
            "  inflating: data/no78_xvid.avi      \n",
            "  inflating: data/no79_xvid.avi      \n",
            "  inflating: data/no7_xvid.avi       \n",
            "  inflating: data/no80_xvid.avi      \n",
            "  inflating: data/no81_xvid.avi      \n",
            "  inflating: data/no82_xvid.avi      \n",
            "  inflating: data/no83_xvid.avi      \n",
            "  inflating: data/no84_xvid.avi      \n",
            "  inflating: data/no85_xvid.avi      \n",
            "  inflating: data/no86_xvid.avi      \n",
            "  inflating: data/no87_xvid.avi      \n",
            "  inflating: data/no88_xvid.avi      \n",
            "  inflating: data/no89_xvid.avi      \n",
            "  inflating: data/no8_xvid.avi       \n",
            "  inflating: data/no90_xvid.avi      \n",
            "  inflating: data/no91_xvid.avi      \n",
            "  inflating: data/no92_xvid.avi      \n",
            "  inflating: data/no93_xvid.avi      \n",
            "  inflating: data/no94_xvid.avi      \n",
            "  inflating: data/no95_xvid.avi      \n",
            "  inflating: data/no96_xvid.avi      \n",
            "  inflating: data/no97_xvid.avi      \n",
            "  inflating: data/no98_xvid.avi      \n",
            "  inflating: data/no99_xvid.avi      \n",
            "  inflating: data/no9_xvid.avi       \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_T5TKsiK5T-q",
        "outputId": "4631e19e-ff6d-4acc-fb58-685312ad01c9"
      },
      "source": [
        "cd data"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/data\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AqkuQ8Tf5WBx"
      },
      "source": [
        "in_dir = \"/content/data\""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4zE-S48pIGzJ"
      },
      "source": [
        "# Frame size  \n",
        "img_size = 224\n",
        "\n",
        "img_size_touple = (img_size, img_size)\n",
        "\n",
        "# Number of channels (RGB)\n",
        "num_channels = 3\n",
        "\n",
        "# Flat frame size\n",
        "img_size_flat = img_size * img_size * num_channels\n",
        "\n",
        "# Number of classes for classification (Violence-No Violence)\n",
        "num_classes = 2\n",
        "\n",
        "# Number of files to train\n",
        "_num_files_train = 1\n",
        "\n",
        "# Number of frames per video\n",
        "_images_per_file = 20\n",
        "\n",
        "# Number of frames per training set\n",
        "_num_images_train = _num_files_train * _images_per_file\n",
        "\n",
        "# Video extension\n",
        "video_exts = \".avi\""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bJTcv1h_Ib5q"
      },
      "source": [
        "def get_frames(current_dir, file_name):\n",
        "    \n",
        "    in_file = os.path.join(current_dir, file_name)\n",
        "    \n",
        "    images = []\n",
        "    \n",
        "    vidcap = cv2.VideoCapture(in_file)\n",
        "    \n",
        "    success,image = vidcap.read()\n",
        "        \n",
        "    count = 0\n",
        "\n",
        "    while count<_images_per_file:\n",
        "                \n",
        "        RGB_img = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n",
        "    \n",
        "        res = cv2.resize(RGB_img, dsize=(img_size, img_size),\n",
        "                                 interpolation=cv2.INTER_CUBIC)\n",
        "    \n",
        "        images.append(res)\n",
        "    \n",
        "        success,image = vidcap.read()\n",
        "    \n",
        "        count += 1\n",
        "        \n",
        "    resul = np.array(images)\n",
        "    \n",
        "    resul = (resul / 255.).astype(np.float16)\n",
        "        \n",
        "    return resul"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Pi7XF9K2IhvB"
      },
      "source": [
        "def label_video_names(in_dir):\n",
        "    \n",
        "    # list containing video names\n",
        "    names = []\n",
        "    # list containin video labels [1, 0] if it has violence and [0, 1] if not\n",
        "    labels = []\n",
        "    \n",
        "    \n",
        "    for current_dir, dir_names,file_names in os.walk(in_dir):\n",
        "        \n",
        "        for file_name in file_names:\n",
        "            \n",
        "            if file_name[0:2] == 'fi':\n",
        "                labels.append([1,0])\n",
        "                names.append(file_name)\n",
        "            elif file_name[0:2] == 'no':\n",
        "                labels.append([0,1])\n",
        "                names.append(file_name)\n",
        "                     \n",
        "            \n",
        "    c = list(zip(names,labels))\n",
        "    # Suffle the data (names and labels)\n",
        "    shuffle(c)\n",
        "    \n",
        "    names, labels = zip(*c)\n",
        "            \n",
        "    return names, labels"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "534c1JbDInrp"
      },
      "source": [
        "# First get the names and labels of the whole videos\n",
        "names, labels = label_video_names(in_dir)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5QyK0hpjKPMV",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "25231575-bffe-43ce-8238-d6c56b993554"
      },
      "source": [
        "names[12]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'no395_xvid.avi'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LscCSUgxKQdK"
      },
      "source": [
        "frames = get_frames(in_dir, names[12])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cDTWK3WrKZPx"
      },
      "source": [
        "visible_frame = (frames*255).astype('uint8')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "33xQh5d0KasJ",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        },
        "outputId": "e5f84296-b689-458f-e75a-8ddd264b1777"
      },
      "source": [
        "plt.imshow(visible_frame[3])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f701d728850>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 32
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PsU1Sk3gKjZ8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        },
        "outputId": "5bd5f2fc-f1c7-43c6-b09d-96e0e39e0d89"
      },
      "source": [
        "plt.imshow(visible_frame[10])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f701d210710>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 33
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EqbsALoeKktx",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "bd02b74d-3e2f-4503-95ef-d52837a115d4"
      },
      "source": [
        "image_model = VGG16(include_top=True, weights='imagenet')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/vgg16/vgg16_weights_tf_dim_ordering_tf_kernels.h5\n",
            "553467904/553467096 [==============================] - 5s 0us/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yqW2aIMqKrXh",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "9c9fbc54-e05d-462c-c645-8b2c286f79ab"
      },
      "source": [
        "image_model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"vgg16\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_1 (InputLayer)         [(None, 224, 224, 3)]     0         \n",
            "_________________________________________________________________\n",
            "block1_conv1 (Conv2D)        (None, 224, 224, 64)      1792      \n",
            "_________________________________________________________________\n",
            "block1_conv2 (Conv2D)        (None, 224, 224, 64)      36928     \n",
            "_________________________________________________________________\n",
            "block1_pool (MaxPooling2D)   (None, 112, 112, 64)      0         \n",
            "_________________________________________________________________\n",
            "block2_conv1 (Conv2D)        (None, 112, 112, 128)     73856     \n",
            "_________________________________________________________________\n",
            "block2_conv2 (Conv2D)        (None, 112, 112, 128)     147584    \n",
            "_________________________________________________________________\n",
            "block2_pool (MaxPooling2D)   (None, 56, 56, 128)       0         \n",
            "_________________________________________________________________\n",
            "block3_conv1 (Conv2D)        (None, 56, 56, 256)       295168    \n",
            "_________________________________________________________________\n",
            "block3_conv2 (Conv2D)        (None, 56, 56, 256)       590080    \n",
            "_________________________________________________________________\n",
            "block3_conv3 (Conv2D)        (None, 56, 56, 256)       590080    \n",
            "_________________________________________________________________\n",
            "block3_pool (MaxPooling2D)   (None, 28, 28, 256)       0         \n",
            "_________________________________________________________________\n",
            "block4_conv1 (Conv2D)        (None, 28, 28, 512)       1180160   \n",
            "_________________________________________________________________\n",
            "block4_conv2 (Conv2D)        (None, 28, 28, 512)       2359808   \n",
            "_________________________________________________________________\n",
            "block4_conv3 (Conv2D)        (None, 28, 28, 512)       2359808   \n",
            "_________________________________________________________________\n",
            "block4_pool (MaxPooling2D)   (None, 14, 14, 512)       0         \n",
            "_________________________________________________________________\n",
            "block5_conv1 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
            "_________________________________________________________________\n",
            "block5_conv2 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
            "_________________________________________________________________\n",
            "block5_conv3 (Conv2D)        (None, 14, 14, 512)       2359808   \n",
            "_________________________________________________________________\n",
            "block5_pool (MaxPooling2D)   (None, 7, 7, 512)         0         \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 25088)             0         \n",
            "_________________________________________________________________\n",
            "fc1 (Dense)                  (None, 4096)              102764544 \n",
            "_________________________________________________________________\n",
            "fc2 (Dense)                  (None, 4096)              16781312  \n",
            "_________________________________________________________________\n",
            "predictions (Dense)          (None, 1000)              4097000   \n",
            "=================================================================\n",
            "Total params: 138,357,544\n",
            "Trainable params: 138,357,544\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "p1TSsXRYK95J",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "1addce16-b22d-47c2-d81c-81600b113380"
      },
      "source": [
        "transfer_layer = image_model.get_layer('fc2')\n",
        "image_model_transfer = Model(inputs=image_model.input, \n",
        "                             outputs=transfer_layer.output)\n",
        "\n",
        "transfer_values_size = K.int_shape(transfer_layer.output)[1]\n",
        "\n",
        "\n",
        "print(\"The input of the VGG16 net have dimensions:\",K.int_shape(image_model.input)[1:3])\n",
        "\n",
        "print(\"The output of the selecter layer of VGG16 net have dimensions: \", transfer_values_size)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "The input of the VGG16 net have dimensions: (224, 224)\n",
            "The output of the selecter layer of VGG16 net have dimensions:  4096\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tHZB2i6jyrwq"
      },
      "source": [
        "def get_vgg16_model():\n",
        "  image_model = VGG16(include_top=True, weights='imagenet')\n",
        "  transfer_layer = image_model.get_layer('fc2')\n",
        "  vgg16_model = Model(inputs=image_model.input, \n",
        "                              outputs=transfer_layer.output)\n",
        "  \n",
        "  return vgg16_model"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZlSeqf2OLFMQ"
      },
      "source": [
        "def get_transfer_values(current_dir, file_name):\n",
        "    \n",
        "    # Pre-allocate input-batch-array for images.\n",
        "    shape = (_images_per_file,) + img_size_touple + (3,)\n",
        "    \n",
        "    image_batch = np.zeros(shape=shape, dtype=np.float16)\n",
        "    \n",
        "    image_batch = get_frames(current_dir, file_name)\n",
        "      \n",
        "    # Pre-allocate output-array for transfer-values.\n",
        "    # Note that we use 16-bit floating-points to save memory.\n",
        "    shape = (_images_per_file, transfer_values_size)\n",
        "    transfer_values = np.zeros(shape=shape, dtype=np.float16)\n",
        "\n",
        "    transfer_values = \\\n",
        "            image_model_transfer.predict(image_batch)\n",
        "            \n",
        "    return transfer_values"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Sidh42bxLLm5"
      },
      "source": [
        "def proces_transfer(vid_names, in_dir, labels):\n",
        "    \n",
        "    count = 0\n",
        "    \n",
        "    tam = len(vid_names)\n",
        "    \n",
        "    # Pre-allocate input-batch-array for images.\n",
        "    shape = (_images_per_file,) + img_size_touple + (3,)\n",
        "    \n",
        "    while count<tam:\n",
        "        \n",
        "        video_name = vid_names[count]\n",
        "        \n",
        "        image_batch = np.zeros(shape=shape, dtype=np.float16)\n",
        "    \n",
        "        image_batch = get_frames(in_dir, video_name)\n",
        "        \n",
        "         # Note that we use 16-bit floating-points to save memory.\n",
        "        shape = (_images_per_file, transfer_values_size)\n",
        "        transfer_values = np.zeros(shape=shape, dtype=np.float16)\n",
        "        \n",
        "        transfer_values = \\\n",
        "            image_model_transfer.predict(image_batch)\n",
        "         \n",
        "        labels1 = labels[count]\n",
        "        \n",
        "        aux = np.ones([20,2])\n",
        "        \n",
        "        labelss = labels1*aux\n",
        "        \n",
        "        yield transfer_values, labelss\n",
        "        \n",
        "        count+=1"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ioVGWt4QLVJ5"
      },
      "source": [
        "def make_files(n_files):\n",
        "    \n",
        "    gen = proces_transfer(names_training, in_dir, labels_training)\n",
        "\n",
        "    numer = 1\n",
        "\n",
        "    # Read the first chunk to get the column dtypes\n",
        "    chunk = next(gen)\n",
        "\n",
        "    row_count = chunk[0].shape[0]\n",
        "    row_count2 = chunk[1].shape[0]\n",
        "    \n",
        "    with h5py.File('prueba.h5', 'w') as f:\n",
        "    \n",
        "        # Initialize a resizable dataset to hold the output\n",
        "        maxshape = (None,) + chunk[0].shape[1:]\n",
        "        maxshape2 = (None,) + chunk[1].shape[1:]\n",
        "    \n",
        "    \n",
        "        dset = f.create_dataset('data', shape=chunk[0].shape, maxshape=maxshape,\n",
        "                                chunks=chunk[0].shape, dtype=chunk[0].dtype)\n",
        "    \n",
        "        dset2 = f.create_dataset('labels', shape=chunk[1].shape, maxshape=maxshape2,\n",
        "                                 chunks=chunk[1].shape, dtype=chunk[1].dtype)\n",
        "    \n",
        "         # Write the first chunk of rows\n",
        "        dset[:] = chunk[0]\n",
        "        dset2[:] = chunk[1]\n",
        "\n",
        "        for chunk in gen:\n",
        "            \n",
        "            if numer == n_files:\n",
        "            \n",
        "                break\n",
        "\n",
        "            # Resize the dataset to accommodate the next chunk of rows\n",
        "            dset.resize(row_count + chunk[0].shape[0], axis=0)\n",
        "            dset2.resize(row_count2 + chunk[1].shape[0], axis=0)\n",
        "\n",
        "            # Write the next chunk\n",
        "            dset[row_count:] = chunk[0]\n",
        "            dset2[row_count:] = chunk[1]\n",
        "\n",
        "            # Increment the row count\n",
        "            row_count += chunk[0].shape[0]\n",
        "            row_count2 += chunk[1].shape[0]\n",
        "            \n",
        "            print_progress(numer, n_files)\n",
        "        \n",
        "            numer += 1"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "l0MyTrBjLXbg"
      },
      "source": [
        "def make_files_test(n_files):\n",
        "    \n",
        "    gen = proces_transfer(names_test, in_dir, labels_test)\n",
        "\n",
        "    numer = 1\n",
        "\n",
        "    # Read the first chunk to get the column dtypes\n",
        "    chunk = next(gen)\n",
        "\n",
        "    row_count = chunk[0].shape[0]\n",
        "    row_count2 = chunk[1].shape[0]\n",
        "    \n",
        "    with h5py.File('pruebavalidation.h5', 'w') as f:\n",
        "    \n",
        "        # Initialize a resizable dataset to hold the output\n",
        "        maxshape = (None,) + chunk[0].shape[1:]\n",
        "        maxshape2 = (None,) + chunk[1].shape[1:]\n",
        "    \n",
        "    \n",
        "        dset = f.create_dataset('data', shape=chunk[0].shape, maxshape=maxshape,\n",
        "                                chunks=chunk[0].shape, dtype=chunk[0].dtype)\n",
        "    \n",
        "        dset2 = f.create_dataset('labels', shape=chunk[1].shape, maxshape=maxshape2,\n",
        "                                 chunks=chunk[1].shape, dtype=chunk[1].dtype)\n",
        "    \n",
        "         # Write the first chunk of rows\n",
        "        dset[:] = chunk[0]\n",
        "        dset2[:] = chunk[1]\n",
        "\n",
        "        for chunk in gen:\n",
        "            \n",
        "            if numer == n_files:\n",
        "            \n",
        "                break\n",
        "\n",
        "            # Resize the dataset to accommodate the next chunk of rows\n",
        "            dset.resize(row_count + chunk[0].shape[0], axis=0)\n",
        "            dset2.resize(row_count2 + chunk[1].shape[0], axis=0)\n",
        "\n",
        "            # Write the next chunk\n",
        "            dset[row_count:] = chunk[0]\n",
        "            dset2[row_count:] = chunk[1]\n",
        "\n",
        "            # Increment the row count\n",
        "            row_count += chunk[0].shape[0]\n",
        "            row_count2 += chunk[1].shape[0]\n",
        "            \n",
        "            print_progress(numer, n_files)\n",
        "        \n",
        "            numer += 1"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0fmFXpUYLaoQ"
      },
      "source": [
        "training_set = int(len(names)*0.8)\n",
        "test_set = int(len(names)*0.2)\n",
        "\n",
        "names_training = names[0:training_set]\n",
        "names_test = names[training_set:]\n",
        "\n",
        "labels_training = labels[0:training_set]\n",
        "labels_test = labels[training_set:]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4t3VaIGXLdWx",
        "outputId": "4fa1389f-ea5e-4982-d7b5-7cdb3c12d95a"
      },
      "source": [
        "make_files(training_set)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "- Progress: 99.9%"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rqI76DChLnko",
        "outputId": "2dbd3064-540c-4833-c716-192ac8a7d2f3"
      },
      "source": [
        "make_files_test(test_set)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "- Progress: 99.5%"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0FQn6AlnPyKS"
      },
      "source": [
        "def process_alldata_training():\n",
        "    \n",
        "    joint_transfer=[]\n",
        "    frames_num=20\n",
        "    count = 0\n",
        "    \n",
        "    with h5py.File('prueba.h5', 'r') as f:\n",
        "            \n",
        "        X_batch = f['data'][:]\n",
        "        y_batch = f['labels'][:]\n",
        "\n",
        "    for i in range(int(len(X_batch)/frames_num)):\n",
        "        inc = count+frames_num\n",
        "        joint_transfer.append([X_batch[count:inc],y_batch[count]])\n",
        "        count =inc\n",
        "        \n",
        "    data =[]\n",
        "    target=[]\n",
        "    \n",
        "    for i in joint_transfer:\n",
        "        data.append(i[0])\n",
        "        target.append(np.array(i[1]))\n",
        "        \n",
        "    return data, target"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "e_SXDzE-P0nN"
      },
      "source": [
        "def process_alldata_test():\n",
        "    \n",
        "    joint_transfer=[]\n",
        "    frames_num=20\n",
        "    count = 0\n",
        "    \n",
        "    with h5py.File('pruebavalidation.h5', 'r') as f:\n",
        "            \n",
        "        X_batch = f['data'][:]\n",
        "        y_batch = f['labels'][:]\n",
        "\n",
        "    for i in range(int(len(X_batch)/frames_num)):\n",
        "        inc = count+frames_num\n",
        "        joint_transfer.append([X_batch[count:inc],y_batch[count]])\n",
        "        count =inc\n",
        "        \n",
        "    data =[]\n",
        "    target=[]\n",
        "    \n",
        "    for i in joint_transfer:\n",
        "        data.append(i[0])\n",
        "        target.append(np.array(i[1]))\n",
        "        \n",
        "    return data, target"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hpoBmKzwP3AZ"
      },
      "source": [
        "data, target = process_alldata_training()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "43zyZmH_P5V6"
      },
      "source": [
        "data_test, target_test = process_alldata_test()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "glmbxANHQPFz"
      },
      "source": [
        "chunk_size = 4096\n",
        "n_chunks = 20\n",
        "rnn_size = 512\n",
        "\n",
        "model = Sequential()\n",
        "model.add(LSTM(rnn_size, input_shape=(n_chunks, chunk_size)))\n",
        "model.add(Dense(1024))\n",
        "model.add(Activation('relu'))\n",
        "model.add(Dense(50))\n",
        "model.add(Activation('sigmoid'))\n",
        "model.add(Dense(2))\n",
        "model.add(Activation('softmax'))\n",
        "model.compile(loss='mean_squared_error', optimizer='adam',metrics=['accuracy'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3Em4lGwpyHmW"
      },
      "source": [
        "def get_lstm_model():  \n",
        "  chunk_size = 4096\n",
        "  n_chunks = 20\n",
        "  rnn_size = 512\n",
        "\n",
        "  model = Sequential()\n",
        "  model.add(LSTM(rnn_size, input_shape=(n_chunks, chunk_size)))\n",
        "  model.add(Dense(1024))\n",
        "  model.add(Activation('relu'))\n",
        "  model.add(Dense(50))\n",
        "  model.add(Activation('sigmoid'))\n",
        "  model.add(Dense(2))\n",
        "  model.add(Activation('softmax'))\n",
        "\n",
        "  return model"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7iIHkiD9rGLx",
        "outputId": "69e988e4-6831-4d19-a519-e32a0b8aed3c"
      },
      "source": [
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "lstm (LSTM)                  (None, 512)               9439232   \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 1024)              525312    \n",
            "_________________________________________________________________\n",
            "activation (Activation)      (None, 1024)              0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 50)                51250     \n",
            "_________________________________________________________________\n",
            "activation_1 (Activation)    (None, 50)                0         \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 2)                 102       \n",
            "_________________________________________________________________\n",
            "activation_2 (Activation)    (None, 2)                 0         \n",
            "=================================================================\n",
            "Total params: 10,015,896\n",
            "Trainable params: 10,015,896\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0N_r9DqdQR6I",
        "outputId": "803d4acd-7197-4e44-8510-e39d8597c29b"
      },
      "source": [
        "epoch = 200\n",
        "batchS = 500\n",
        "\n",
        "model.fit(np.array(data[0:750]), np.array(target[0:750]), epochs=epoch,\n",
        "                    validation_data=(np.array(data[750:]), np.array(target[750:])), \n",
        "                    batch_size=batchS, verbose=2)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "2/2 - 20s - loss: 0.3499 - accuracy: 0.4773 - val_loss: 0.3292 - val_accuracy: 0.4600\n",
            "Epoch 2/200\n",
            "2/2 - 18s - loss: 0.2901 - accuracy: 0.5013 - val_loss: 0.2619 - val_accuracy: 0.5400\n",
            "Epoch 3/200\n",
            "2/2 - 13s - loss: 0.2693 - accuracy: 0.5013 - val_loss: 0.2481 - val_accuracy: 0.4600\n",
            "Epoch 4/200\n",
            "2/2 - 12s - loss: 0.2511 - accuracy: 0.4987 - val_loss: 0.2640 - val_accuracy: 0.4600\n",
            "Epoch 5/200\n",
            "2/2 - 11s - loss: 0.2547 - accuracy: 0.4987 - val_loss: 0.2462 - val_accuracy: 0.9400\n",
            "Epoch 6/200\n",
            "2/2 - 12s - loss: 0.2490 - accuracy: 0.7227 - val_loss: 0.2474 - val_accuracy: 0.5400\n",
            "Epoch 7/200\n",
            "2/2 - 12s - loss: 0.2551 - accuracy: 0.5013 - val_loss: 0.2418 - val_accuracy: 0.5400\n",
            "Epoch 8/200\n",
            "2/2 - 11s - loss: 0.2451 - accuracy: 0.5480 - val_loss: 0.2475 - val_accuracy: 0.4600\n",
            "Epoch 9/200\n",
            "2/2 - 12s - loss: 0.2450 - accuracy: 0.4987 - val_loss: 0.2433 - val_accuracy: 0.4600\n",
            "Epoch 10/200\n",
            "2/2 - 11s - loss: 0.2402 - accuracy: 0.5000 - val_loss: 0.2305 - val_accuracy: 0.8800\n",
            "Epoch 11/200\n",
            "2/2 - 12s - loss: 0.2347 - accuracy: 0.7133 - val_loss: 0.2208 - val_accuracy: 0.5800\n",
            "Epoch 12/200\n",
            "2/2 - 11s - loss: 0.2280 - accuracy: 0.6413 - val_loss: 0.2055 - val_accuracy: 0.9600\n",
            "Epoch 13/200\n",
            "2/2 - 11s - loss: 0.2120 - accuracy: 0.8920 - val_loss: 0.1823 - val_accuracy: 0.9600\n",
            "Epoch 14/200\n",
            "2/2 - 12s - loss: 0.1938 - accuracy: 0.8827 - val_loss: 0.1561 - val_accuracy: 0.9600\n",
            "Epoch 15/200\n",
            "2/2 - 12s - loss: 0.1721 - accuracy: 0.8947 - val_loss: 0.1234 - val_accuracy: 0.9600\n",
            "Epoch 16/200\n",
            "2/2 - 11s - loss: 0.1467 - accuracy: 0.8907 - val_loss: 0.0905 - val_accuracy: 0.9600\n",
            "Epoch 17/200\n",
            "2/2 - 11s - loss: 0.1215 - accuracy: 0.8960 - val_loss: 0.0672 - val_accuracy: 0.9600\n",
            "Epoch 18/200\n",
            "2/2 - 11s - loss: 0.1054 - accuracy: 0.8853 - val_loss: 0.0556 - val_accuracy: 0.9400\n",
            "Epoch 19/200\n",
            "2/2 - 12s - loss: 0.1024 - accuracy: 0.8733 - val_loss: 0.0446 - val_accuracy: 0.9600\n",
            "Epoch 20/200\n",
            "2/2 - 12s - loss: 0.0926 - accuracy: 0.8800 - val_loss: 0.0486 - val_accuracy: 0.9600\n",
            "Epoch 21/200\n",
            "2/2 - 11s - loss: 0.0938 - accuracy: 0.8827 - val_loss: 0.0410 - val_accuracy: 0.9400\n",
            "Epoch 22/200\n",
            "2/2 - 11s - loss: 0.0955 - accuracy: 0.8733 - val_loss: 0.0379 - val_accuracy: 0.9600\n",
            "Epoch 23/200\n",
            "2/2 - 11s - loss: 0.0893 - accuracy: 0.8827 - val_loss: 0.0416 - val_accuracy: 0.9600\n",
            "Epoch 24/200\n",
            "2/2 - 11s - loss: 0.0866 - accuracy: 0.8947 - val_loss: 0.0447 - val_accuracy: 0.9400\n",
            "Epoch 25/200\n",
            "2/2 - 11s - loss: 0.0954 - accuracy: 0.8747 - val_loss: 0.0401 - val_accuracy: 0.9600\n",
            "Epoch 26/200\n",
            "2/2 - 11s - loss: 0.0859 - accuracy: 0.8947 - val_loss: 0.0389 - val_accuracy: 0.9600\n",
            "Epoch 27/200\n",
            "2/2 - 12s - loss: 0.0813 - accuracy: 0.9040 - val_loss: 0.0402 - val_accuracy: 0.9600\n",
            "Epoch 28/200\n",
            "2/2 - 12s - loss: 0.0828 - accuracy: 0.8933 - val_loss: 0.0420 - val_accuracy: 0.9600\n",
            "Epoch 29/200\n",
            "2/2 - 12s - loss: 0.0837 - accuracy: 0.9000 - val_loss: 0.0390 - val_accuracy: 0.9600\n",
            "Epoch 30/200\n",
            "2/2 - 12s - loss: 0.0783 - accuracy: 0.9027 - val_loss: 0.0394 - val_accuracy: 0.9600\n",
            "Epoch 31/200\n",
            "2/2 - 11s - loss: 0.0763 - accuracy: 0.9040 - val_loss: 0.0412 - val_accuracy: 0.9600\n",
            "Epoch 32/200\n",
            "2/2 - 12s - loss: 0.0779 - accuracy: 0.9080 - val_loss: 0.0413 - val_accuracy: 0.9600\n",
            "Epoch 33/200\n",
            "2/2 - 12s - loss: 0.0790 - accuracy: 0.8987 - val_loss: 0.0408 - val_accuracy: 0.9600\n",
            "Epoch 34/200\n",
            "2/2 - 12s - loss: 0.0731 - accuracy: 0.9107 - val_loss: 0.0442 - val_accuracy: 0.9600\n",
            "Epoch 35/200\n",
            "2/2 - 12s - loss: 0.0819 - accuracy: 0.8960 - val_loss: 0.0426 - val_accuracy: 0.9600\n",
            "Epoch 36/200\n",
            "2/2 - 12s - loss: 0.0787 - accuracy: 0.8987 - val_loss: 0.0406 - val_accuracy: 0.9600\n",
            "Epoch 37/200\n",
            "2/2 - 11s - loss: 0.0713 - accuracy: 0.9107 - val_loss: 0.0437 - val_accuracy: 0.9600\n",
            "Epoch 38/200\n",
            "2/2 - 12s - loss: 0.0805 - accuracy: 0.8960 - val_loss: 0.0444 - val_accuracy: 0.9600\n",
            "Epoch 39/200\n",
            "2/2 - 12s - loss: 0.0757 - accuracy: 0.8987 - val_loss: 0.0403 - val_accuracy: 0.9600\n",
            "Epoch 40/200\n",
            "2/2 - 12s - loss: 0.0701 - accuracy: 0.9173 - val_loss: 0.0413 - val_accuracy: 0.9600\n",
            "Epoch 41/200\n",
            "2/2 - 12s - loss: 0.0701 - accuracy: 0.9173 - val_loss: 0.0457 - val_accuracy: 0.9400\n",
            "Epoch 42/200\n",
            "2/2 - 11s - loss: 0.0749 - accuracy: 0.9013 - val_loss: 0.0412 - val_accuracy: 0.9600\n",
            "Epoch 43/200\n",
            "2/2 - 11s - loss: 0.0673 - accuracy: 0.9200 - val_loss: 0.0424 - val_accuracy: 0.9600\n",
            "Epoch 44/200\n",
            "2/2 - 12s - loss: 0.0708 - accuracy: 0.9160 - val_loss: 0.0468 - val_accuracy: 0.9400\n",
            "Epoch 45/200\n",
            "2/2 - 12s - loss: 0.0752 - accuracy: 0.9053 - val_loss: 0.0410 - val_accuracy: 0.9600\n",
            "Epoch 46/200\n",
            "2/2 - 11s - loss: 0.0685 - accuracy: 0.9213 - val_loss: 0.0413 - val_accuracy: 0.9600\n",
            "Epoch 47/200\n",
            "2/2 - 11s - loss: 0.0707 - accuracy: 0.9147 - val_loss: 0.0446 - val_accuracy: 0.9600\n",
            "Epoch 48/200\n",
            "2/2 - 11s - loss: 0.0675 - accuracy: 0.9120 - val_loss: 0.0415 - val_accuracy: 0.9600\n",
            "Epoch 49/200\n",
            "2/2 - 11s - loss: 0.0680 - accuracy: 0.9213 - val_loss: 0.0421 - val_accuracy: 0.9600\n",
            "Epoch 50/200\n",
            "2/2 - 11s - loss: 0.0650 - accuracy: 0.9213 - val_loss: 0.0426 - val_accuracy: 0.9600\n",
            "Epoch 51/200\n",
            "2/2 - 12s - loss: 0.0644 - accuracy: 0.9147 - val_loss: 0.0408 - val_accuracy: 0.9600\n",
            "Epoch 52/200\n",
            "2/2 - 11s - loss: 0.0638 - accuracy: 0.9267 - val_loss: 0.0419 - val_accuracy: 0.9600\n",
            "Epoch 53/200\n",
            "2/2 - 11s - loss: 0.0627 - accuracy: 0.9253 - val_loss: 0.0405 - val_accuracy: 0.9600\n",
            "Epoch 54/200\n",
            "2/2 - 11s - loss: 0.0622 - accuracy: 0.9293 - val_loss: 0.0409 - val_accuracy: 0.9600\n",
            "Epoch 55/200\n",
            "2/2 - 12s - loss: 0.0598 - accuracy: 0.9307 - val_loss: 0.0450 - val_accuracy: 0.9600\n",
            "Epoch 56/200\n",
            "2/2 - 11s - loss: 0.0625 - accuracy: 0.9227 - val_loss: 0.0411 - val_accuracy: 0.9600\n",
            "Epoch 57/200\n",
            "2/2 - 11s - loss: 0.0619 - accuracy: 0.9293 - val_loss: 0.0430 - val_accuracy: 0.9600\n",
            "Epoch 58/200\n",
            "2/2 - 11s - loss: 0.0617 - accuracy: 0.9267 - val_loss: 0.0406 - val_accuracy: 0.9600\n",
            "Epoch 59/200\n",
            "2/2 - 11s - loss: 0.0598 - accuracy: 0.9307 - val_loss: 0.0408 - val_accuracy: 0.9600\n",
            "Epoch 60/200\n",
            "2/2 - 11s - loss: 0.0586 - accuracy: 0.9253 - val_loss: 0.0421 - val_accuracy: 0.9600\n",
            "Epoch 61/200\n",
            "2/2 - 11s - loss: 0.0596 - accuracy: 0.9347 - val_loss: 0.0410 - val_accuracy: 0.9600\n",
            "Epoch 62/200\n",
            "2/2 - 11s - loss: 0.0580 - accuracy: 0.9320 - val_loss: 0.0405 - val_accuracy: 0.9600\n",
            "Epoch 63/200\n",
            "2/2 - 11s - loss: 0.0612 - accuracy: 0.9307 - val_loss: 0.0435 - val_accuracy: 0.9600\n",
            "Epoch 64/200\n",
            "2/2 - 11s - loss: 0.0607 - accuracy: 0.9293 - val_loss: 0.0404 - val_accuracy: 0.9600\n",
            "Epoch 65/200\n",
            "2/2 - 11s - loss: 0.0578 - accuracy: 0.9347 - val_loss: 0.0411 - val_accuracy: 0.9600\n",
            "Epoch 66/200\n",
            "2/2 - 11s - loss: 0.0587 - accuracy: 0.9333 - val_loss: 0.0407 - val_accuracy: 0.9600\n",
            "Epoch 67/200\n",
            "2/2 - 11s - loss: 0.0550 - accuracy: 0.9347 - val_loss: 0.0404 - val_accuracy: 0.9600\n",
            "Epoch 68/200\n",
            "2/2 - 11s - loss: 0.0545 - accuracy: 0.9360 - val_loss: 0.0461 - val_accuracy: 0.9400\n",
            "Epoch 69/200\n",
            "2/2 - 11s - loss: 0.0591 - accuracy: 0.9280 - val_loss: 0.0450 - val_accuracy: 0.9400\n",
            "Epoch 70/200\n",
            "2/2 - 12s - loss: 0.0617 - accuracy: 0.9267 - val_loss: 0.0583 - val_accuracy: 0.9200\n",
            "Epoch 71/200\n",
            "2/2 - 11s - loss: 0.0716 - accuracy: 0.9120 - val_loss: 0.0586 - val_accuracy: 0.9400\n",
            "Epoch 72/200\n",
            "2/2 - 11s - loss: 0.0887 - accuracy: 0.8827 - val_loss: 0.0595 - val_accuracy: 0.9200\n",
            "Epoch 73/200\n",
            "2/2 - 11s - loss: 0.0838 - accuracy: 0.8960 - val_loss: 0.0404 - val_accuracy: 0.9600\n",
            "Epoch 74/200\n",
            "2/2 - 11s - loss: 0.0689 - accuracy: 0.9133 - val_loss: 0.0407 - val_accuracy: 0.9600\n",
            "Epoch 75/200\n",
            "2/2 - 11s - loss: 0.0631 - accuracy: 0.9213 - val_loss: 0.0451 - val_accuracy: 0.9400\n",
            "Epoch 76/200\n",
            "2/2 - 11s - loss: 0.0589 - accuracy: 0.9307 - val_loss: 0.0453 - val_accuracy: 0.9400\n",
            "Epoch 77/200\n",
            "2/2 - 12s - loss: 0.0662 - accuracy: 0.9173 - val_loss: 0.0417 - val_accuracy: 0.9600\n",
            "Epoch 78/200\n",
            "2/2 - 11s - loss: 0.0565 - accuracy: 0.9307 - val_loss: 0.0396 - val_accuracy: 0.9600\n",
            "Epoch 79/200\n",
            "2/2 - 11s - loss: 0.0556 - accuracy: 0.9373 - val_loss: 0.0396 - val_accuracy: 0.9600\n",
            "Epoch 80/200\n",
            "2/2 - 11s - loss: 0.0518 - accuracy: 0.9413 - val_loss: 0.0419 - val_accuracy: 0.9600\n",
            "Epoch 81/200\n",
            "2/2 - 11s - loss: 0.0552 - accuracy: 0.9293 - val_loss: 0.0410 - val_accuracy: 0.9600\n",
            "Epoch 82/200\n",
            "2/2 - 11s - loss: 0.0555 - accuracy: 0.9387 - val_loss: 0.0398 - val_accuracy: 0.9600\n",
            "Epoch 83/200\n",
            "2/2 - 11s - loss: 0.0509 - accuracy: 0.9453 - val_loss: 0.0419 - val_accuracy: 0.9600\n",
            "Epoch 84/200\n",
            "2/2 - 11s - loss: 0.0549 - accuracy: 0.9293 - val_loss: 0.0399 - val_accuracy: 0.9600\n",
            "Epoch 85/200\n",
            "2/2 - 11s - loss: 0.0537 - accuracy: 0.9400 - val_loss: 0.0403 - val_accuracy: 0.9600\n",
            "Epoch 86/200\n",
            "2/2 - 11s - loss: 0.0522 - accuracy: 0.9360 - val_loss: 0.0418 - val_accuracy: 0.9600\n",
            "Epoch 87/200\n",
            "2/2 - 11s - loss: 0.0512 - accuracy: 0.9373 - val_loss: 0.0447 - val_accuracy: 0.9400\n",
            "Epoch 88/200\n",
            "2/2 - 11s - loss: 0.0587 - accuracy: 0.9320 - val_loss: 0.0415 - val_accuracy: 0.9600\n",
            "Epoch 89/200\n",
            "2/2 - 11s - loss: 0.0543 - accuracy: 0.9333 - val_loss: 0.0401 - val_accuracy: 0.9600\n",
            "Epoch 90/200\n",
            "2/2 - 11s - loss: 0.0501 - accuracy: 0.9480 - val_loss: 0.0402 - val_accuracy: 0.9600\n",
            "Epoch 91/200\n",
            "2/2 - 11s - loss: 0.0497 - accuracy: 0.9453 - val_loss: 0.0411 - val_accuracy: 0.9600\n",
            "Epoch 92/200\n",
            "2/2 - 11s - loss: 0.0499 - accuracy: 0.9373 - val_loss: 0.0435 - val_accuracy: 0.9600\n",
            "Epoch 93/200\n",
            "2/2 - 11s - loss: 0.0549 - accuracy: 0.9413 - val_loss: 0.0439 - val_accuracy: 0.9400\n",
            "Epoch 94/200\n",
            "2/2 - 11s - loss: 0.0577 - accuracy: 0.9307 - val_loss: 0.0398 - val_accuracy: 0.9600\n",
            "Epoch 95/200\n",
            "2/2 - 12s - loss: 0.0489 - accuracy: 0.9480 - val_loss: 0.0407 - val_accuracy: 0.9600\n",
            "Epoch 96/200\n",
            "2/2 - 11s - loss: 0.0464 - accuracy: 0.9520 - val_loss: 0.0465 - val_accuracy: 0.9400\n",
            "Epoch 97/200\n",
            "2/2 - 11s - loss: 0.0579 - accuracy: 0.9293 - val_loss: 0.0410 - val_accuracy: 0.9600\n",
            "Epoch 98/200\n",
            "2/2 - 11s - loss: 0.0564 - accuracy: 0.9360 - val_loss: 0.0464 - val_accuracy: 0.9400\n",
            "Epoch 99/200\n",
            "2/2 - 11s - loss: 0.0645 - accuracy: 0.9187 - val_loss: 0.0404 - val_accuracy: 0.9600\n",
            "Epoch 100/200\n",
            "2/2 - 11s - loss: 0.0568 - accuracy: 0.9253 - val_loss: 0.0416 - val_accuracy: 0.9600\n",
            "Epoch 101/200\n",
            "2/2 - 11s - loss: 0.0507 - accuracy: 0.9427 - val_loss: 0.0447 - val_accuracy: 0.9400\n",
            "Epoch 102/200\n",
            "2/2 - 11s - loss: 0.0557 - accuracy: 0.9320 - val_loss: 0.0423 - val_accuracy: 0.9600\n",
            "Epoch 103/200\n",
            "2/2 - 11s - loss: 0.0520 - accuracy: 0.9440 - val_loss: 0.0405 - val_accuracy: 0.9600\n",
            "Epoch 104/200\n",
            "2/2 - 11s - loss: 0.0489 - accuracy: 0.9400 - val_loss: 0.0394 - val_accuracy: 0.9600\n",
            "Epoch 105/200\n",
            "2/2 - 11s - loss: 0.0471 - accuracy: 0.9493 - val_loss: 0.0398 - val_accuracy: 0.9600\n",
            "Epoch 106/200\n",
            "2/2 - 11s - loss: 0.0457 - accuracy: 0.9507 - val_loss: 0.0406 - val_accuracy: 0.9600\n",
            "Epoch 107/200\n",
            "2/2 - 11s - loss: 0.0481 - accuracy: 0.9400 - val_loss: 0.0395 - val_accuracy: 0.9600\n",
            "Epoch 108/200\n",
            "2/2 - 11s - loss: 0.0449 - accuracy: 0.9493 - val_loss: 0.0400 - val_accuracy: 0.9600\n",
            "Epoch 109/200\n",
            "2/2 - 11s - loss: 0.0455 - accuracy: 0.9520 - val_loss: 0.0402 - val_accuracy: 0.9600\n",
            "Epoch 110/200\n",
            "2/2 - 11s - loss: 0.0459 - accuracy: 0.9440 - val_loss: 0.0399 - val_accuracy: 0.9600\n",
            "Epoch 111/200\n",
            "2/2 - 11s - loss: 0.0451 - accuracy: 0.9520 - val_loss: 0.0397 - val_accuracy: 0.9600\n",
            "Epoch 112/200\n",
            "2/2 - 12s - loss: 0.0439 - accuracy: 0.9480 - val_loss: 0.0399 - val_accuracy: 0.9600\n",
            "Epoch 113/200\n",
            "2/2 - 11s - loss: 0.0446 - accuracy: 0.9453 - val_loss: 0.0397 - val_accuracy: 0.9600\n",
            "Epoch 114/200\n",
            "2/2 - 11s - loss: 0.0434 - accuracy: 0.9493 - val_loss: 0.0402 - val_accuracy: 0.9600\n",
            "Epoch 115/200\n",
            "2/2 - 11s - loss: 0.0446 - accuracy: 0.9507 - val_loss: 0.0398 - val_accuracy: 0.9600\n",
            "Epoch 116/200\n",
            "2/2 - 11s - loss: 0.0432 - accuracy: 0.9493 - val_loss: 0.0401 - val_accuracy: 0.9600\n",
            "Epoch 117/200\n",
            "2/2 - 11s - loss: 0.0435 - accuracy: 0.9533 - val_loss: 0.0397 - val_accuracy: 0.9600\n",
            "Epoch 118/200\n",
            "2/2 - 11s - loss: 0.0423 - accuracy: 0.9493 - val_loss: 0.0401 - val_accuracy: 0.9600\n",
            "Epoch 119/200\n",
            "2/2 - 11s - loss: 0.0428 - accuracy: 0.9507 - val_loss: 0.0397 - val_accuracy: 0.9600\n",
            "Epoch 120/200\n",
            "2/2 - 11s - loss: 0.0416 - accuracy: 0.9507 - val_loss: 0.0402 - val_accuracy: 0.9600\n",
            "Epoch 121/200\n",
            "2/2 - 11s - loss: 0.0424 - accuracy: 0.9533 - val_loss: 0.0456 - val_accuracy: 0.9400\n",
            "Epoch 122/200\n",
            "2/2 - 11s - loss: 0.0581 - accuracy: 0.9280 - val_loss: 0.0409 - val_accuracy: 0.9600\n",
            "Epoch 123/200\n",
            "2/2 - 11s - loss: 0.0464 - accuracy: 0.9467 - val_loss: 0.0396 - val_accuracy: 0.9600\n",
            "Epoch 124/200\n",
            "2/2 - 11s - loss: 0.0429 - accuracy: 0.9493 - val_loss: 0.0397 - val_accuracy: 0.9600\n",
            "Epoch 125/200\n",
            "2/2 - 11s - loss: 0.0428 - accuracy: 0.9520 - val_loss: 0.0398 - val_accuracy: 0.9600\n",
            "Epoch 126/200\n",
            "2/2 - 11s - loss: 0.0427 - accuracy: 0.9533 - val_loss: 0.0397 - val_accuracy: 0.9600\n",
            "Epoch 127/200\n",
            "2/2 - 11s - loss: 0.0421 - accuracy: 0.9547 - val_loss: 0.0398 - val_accuracy: 0.9600\n",
            "Epoch 128/200\n",
            "2/2 - 11s - loss: 0.0415 - accuracy: 0.9533 - val_loss: 0.0396 - val_accuracy: 0.9600\n",
            "Epoch 129/200\n",
            "2/2 - 11s - loss: 0.0413 - accuracy: 0.9547 - val_loss: 0.0396 - val_accuracy: 0.9600\n",
            "Epoch 130/200\n",
            "2/2 - 11s - loss: 0.0422 - accuracy: 0.9493 - val_loss: 0.0404 - val_accuracy: 0.9600\n",
            "Epoch 131/200\n",
            "2/2 - 11s - loss: 0.0428 - accuracy: 0.9560 - val_loss: 0.0399 - val_accuracy: 0.9600\n",
            "Epoch 132/200\n",
            "2/2 - 11s - loss: 0.0398 - accuracy: 0.9573 - val_loss: 0.0406 - val_accuracy: 0.9600\n",
            "Epoch 133/200\n",
            "2/2 - 11s - loss: 0.0398 - accuracy: 0.9587 - val_loss: 0.0428 - val_accuracy: 0.9600\n",
            "Epoch 134/200\n",
            "2/2 - 11s - loss: 0.0441 - accuracy: 0.9493 - val_loss: 0.0510 - val_accuracy: 0.9400\n",
            "Epoch 135/200\n",
            "2/2 - 11s - loss: 0.0561 - accuracy: 0.9293 - val_loss: 0.0409 - val_accuracy: 0.9600\n",
            "Epoch 136/200\n",
            "2/2 - 11s - loss: 0.0414 - accuracy: 0.9560 - val_loss: 0.0450 - val_accuracy: 0.9400\n",
            "Epoch 137/200\n",
            "2/2 - 11s - loss: 0.0423 - accuracy: 0.9507 - val_loss: 0.0447 - val_accuracy: 0.9400\n",
            "Epoch 138/200\n",
            "2/2 - 11s - loss: 0.0517 - accuracy: 0.9373 - val_loss: 0.0427 - val_accuracy: 0.9600\n",
            "Epoch 139/200\n",
            "2/2 - 11s - loss: 0.0434 - accuracy: 0.9547 - val_loss: 0.0399 - val_accuracy: 0.9600\n",
            "Epoch 140/200\n",
            "2/2 - 11s - loss: 0.0413 - accuracy: 0.9547 - val_loss: 0.0406 - val_accuracy: 0.9600\n",
            "Epoch 141/200\n",
            "2/2 - 11s - loss: 0.0417 - accuracy: 0.9560 - val_loss: 0.0437 - val_accuracy: 0.9400\n",
            "Epoch 142/200\n",
            "2/2 - 11s - loss: 0.0412 - accuracy: 0.9573 - val_loss: 0.0406 - val_accuracy: 0.9600\n",
            "Epoch 143/200\n",
            "2/2 - 11s - loss: 0.0409 - accuracy: 0.9560 - val_loss: 0.0424 - val_accuracy: 0.9600\n",
            "Epoch 144/200\n",
            "2/2 - 11s - loss: 0.0396 - accuracy: 0.9587 - val_loss: 0.0416 - val_accuracy: 0.9600\n",
            "Epoch 145/200\n",
            "2/2 - 11s - loss: 0.0439 - accuracy: 0.9493 - val_loss: 0.0398 - val_accuracy: 0.9600\n",
            "Epoch 146/200\n",
            "2/2 - 11s - loss: 0.0378 - accuracy: 0.9587 - val_loss: 0.0420 - val_accuracy: 0.9600\n",
            "Epoch 147/200\n",
            "2/2 - 11s - loss: 0.0385 - accuracy: 0.9600 - val_loss: 0.0400 - val_accuracy: 0.9600\n",
            "Epoch 148/200\n",
            "2/2 - 11s - loss: 0.0381 - accuracy: 0.9587 - val_loss: 0.0403 - val_accuracy: 0.9600\n",
            "Epoch 149/200\n",
            "2/2 - 11s - loss: 0.0380 - accuracy: 0.9627 - val_loss: 0.0421 - val_accuracy: 0.9600\n",
            "Epoch 150/200\n",
            "2/2 - 11s - loss: 0.0418 - accuracy: 0.9560 - val_loss: 0.0422 - val_accuracy: 0.9600\n",
            "Epoch 151/200\n",
            "2/2 - 11s - loss: 0.0372 - accuracy: 0.9613 - val_loss: 0.0421 - val_accuracy: 0.9600\n",
            "Epoch 152/200\n",
            "2/2 - 11s - loss: 0.0403 - accuracy: 0.9587 - val_loss: 0.0445 - val_accuracy: 0.9400\n",
            "Epoch 153/200\n",
            "2/2 - 11s - loss: 0.0385 - accuracy: 0.9560 - val_loss: 0.0411 - val_accuracy: 0.9600\n",
            "Epoch 154/200\n",
            "2/2 - 12s - loss: 0.0443 - accuracy: 0.9507 - val_loss: 0.0402 - val_accuracy: 0.9600\n",
            "Epoch 155/200\n",
            "2/2 - 11s - loss: 0.0376 - accuracy: 0.9600 - val_loss: 0.0401 - val_accuracy: 0.9600\n",
            "Epoch 156/200\n",
            "2/2 - 11s - loss: 0.0370 - accuracy: 0.9613 - val_loss: 0.0414 - val_accuracy: 0.9600\n",
            "Epoch 157/200\n",
            "2/2 - 11s - loss: 0.0359 - accuracy: 0.9640 - val_loss: 0.0412 - val_accuracy: 0.9600\n",
            "Epoch 158/200\n",
            "2/2 - 11s - loss: 0.0406 - accuracy: 0.9547 - val_loss: 0.0408 - val_accuracy: 0.9600\n",
            "Epoch 159/200\n",
            "2/2 - 11s - loss: 0.0368 - accuracy: 0.9627 - val_loss: 0.0435 - val_accuracy: 0.9600\n",
            "Epoch 160/200\n",
            "2/2 - 11s - loss: 0.0380 - accuracy: 0.9600 - val_loss: 0.0443 - val_accuracy: 0.9400\n",
            "Epoch 161/200\n",
            "2/2 - 11s - loss: 0.0409 - accuracy: 0.9520 - val_loss: 0.0534 - val_accuracy: 0.9400\n",
            "Epoch 162/200\n",
            "2/2 - 11s - loss: 0.0521 - accuracy: 0.9333 - val_loss: 0.0436 - val_accuracy: 0.9600\n",
            "Epoch 163/200\n",
            "2/2 - 11s - loss: 0.0428 - accuracy: 0.9533 - val_loss: 0.0483 - val_accuracy: 0.9400\n",
            "Epoch 164/200\n",
            "2/2 - 11s - loss: 0.0395 - accuracy: 0.9600 - val_loss: 0.0409 - val_accuracy: 0.9600\n",
            "Epoch 165/200\n",
            "2/2 - 11s - loss: 0.0412 - accuracy: 0.9573 - val_loss: 0.0414 - val_accuracy: 0.9600\n",
            "Epoch 166/200\n",
            "2/2 - 11s - loss: 0.0414 - accuracy: 0.9533 - val_loss: 0.0496 - val_accuracy: 0.9400\n",
            "Epoch 167/200\n",
            "2/2 - 11s - loss: 0.0450 - accuracy: 0.9507 - val_loss: 0.0570 - val_accuracy: 0.9400\n",
            "Epoch 168/200\n",
            "2/2 - 11s - loss: 0.0791 - accuracy: 0.8960 - val_loss: 0.0531 - val_accuracy: 0.9400\n",
            "Epoch 169/200\n",
            "2/2 - 11s - loss: 0.0550 - accuracy: 0.9320 - val_loss: 0.0462 - val_accuracy: 0.9400\n",
            "Epoch 170/200\n",
            "2/2 - 11s - loss: 0.0499 - accuracy: 0.9453 - val_loss: 0.0516 - val_accuracy: 0.9400\n",
            "Epoch 171/200\n",
            "2/2 - 11s - loss: 0.0406 - accuracy: 0.9560 - val_loss: 0.0468 - val_accuracy: 0.9400\n",
            "Epoch 172/200\n",
            "2/2 - 11s - loss: 0.0468 - accuracy: 0.9467 - val_loss: 0.0520 - val_accuracy: 0.9400\n",
            "Epoch 173/200\n",
            "2/2 - 11s - loss: 0.0439 - accuracy: 0.9493 - val_loss: 0.0418 - val_accuracy: 0.9600\n",
            "Epoch 174/200\n",
            "2/2 - 11s - loss: 0.0421 - accuracy: 0.9507 - val_loss: 0.0475 - val_accuracy: 0.9400\n",
            "Epoch 175/200\n",
            "2/2 - 11s - loss: 0.0395 - accuracy: 0.9573 - val_loss: 0.0402 - val_accuracy: 0.9600\n",
            "Epoch 176/200\n",
            "2/2 - 11s - loss: 0.0371 - accuracy: 0.9613 - val_loss: 0.0405 - val_accuracy: 0.9600\n",
            "Epoch 177/200\n",
            "2/2 - 11s - loss: 0.0392 - accuracy: 0.9587 - val_loss: 0.0490 - val_accuracy: 0.9400\n",
            "Epoch 178/200\n",
            "2/2 - 11s - loss: 0.0379 - accuracy: 0.9640 - val_loss: 0.0426 - val_accuracy: 0.9600\n",
            "Epoch 179/200\n",
            "2/2 - 11s - loss: 0.0431 - accuracy: 0.9520 - val_loss: 0.0516 - val_accuracy: 0.9400\n",
            "Epoch 180/200\n",
            "2/2 - 11s - loss: 0.0413 - accuracy: 0.9547 - val_loss: 0.0443 - val_accuracy: 0.9400\n",
            "Epoch 181/200\n",
            "2/2 - 11s - loss: 0.0427 - accuracy: 0.9547 - val_loss: 0.0524 - val_accuracy: 0.9400\n",
            "Epoch 182/200\n",
            "2/2 - 11s - loss: 0.0400 - accuracy: 0.9600 - val_loss: 0.0495 - val_accuracy: 0.9400\n",
            "Epoch 183/200\n",
            "2/2 - 12s - loss: 0.0475 - accuracy: 0.9453 - val_loss: 0.0523 - val_accuracy: 0.9400\n",
            "Epoch 184/200\n",
            "2/2 - 12s - loss: 0.0529 - accuracy: 0.9360 - val_loss: 0.0523 - val_accuracy: 0.9400\n",
            "Epoch 185/200\n",
            "2/2 - 11s - loss: 0.0561 - accuracy: 0.9307 - val_loss: 0.0504 - val_accuracy: 0.9400\n",
            "Epoch 186/200\n",
            "2/2 - 11s - loss: 0.0621 - accuracy: 0.9267 - val_loss: 0.0441 - val_accuracy: 0.9400\n",
            "Epoch 187/200\n",
            "2/2 - 12s - loss: 0.0446 - accuracy: 0.9493 - val_loss: 0.0442 - val_accuracy: 0.9400\n",
            "Epoch 188/200\n",
            "2/2 - 12s - loss: 0.0373 - accuracy: 0.9600 - val_loss: 0.0448 - val_accuracy: 0.9400\n",
            "Epoch 189/200\n",
            "2/2 - 11s - loss: 0.0371 - accuracy: 0.9600 - val_loss: 0.0408 - val_accuracy: 0.9600\n",
            "Epoch 190/200\n",
            "2/2 - 11s - loss: 0.0366 - accuracy: 0.9613 - val_loss: 0.0527 - val_accuracy: 0.9400\n",
            "Epoch 191/200\n",
            "2/2 - 11s - loss: 0.0427 - accuracy: 0.9520 - val_loss: 0.0412 - val_accuracy: 0.9600\n",
            "Epoch 192/200\n",
            "2/2 - 11s - loss: 0.0406 - accuracy: 0.9533 - val_loss: 0.0425 - val_accuracy: 0.9600\n",
            "Epoch 193/200\n",
            "2/2 - 11s - loss: 0.0349 - accuracy: 0.9640 - val_loss: 0.0489 - val_accuracy: 0.9400\n",
            "Epoch 194/200\n",
            "2/2 - 11s - loss: 0.0377 - accuracy: 0.9627 - val_loss: 0.0464 - val_accuracy: 0.9400\n",
            "Epoch 195/200\n",
            "2/2 - 12s - loss: 0.0436 - accuracy: 0.9533 - val_loss: 0.0531 - val_accuracy: 0.9400\n",
            "Epoch 196/200\n",
            "2/2 - 11s - loss: 0.0510 - accuracy: 0.9347 - val_loss: 0.0473 - val_accuracy: 0.9400\n",
            "Epoch 197/200\n",
            "2/2 - 11s - loss: 0.0475 - accuracy: 0.9467 - val_loss: 0.0533 - val_accuracy: 0.9400\n",
            "Epoch 198/200\n",
            "2/2 - 11s - loss: 0.0450 - accuracy: 0.9480 - val_loss: 0.0404 - val_accuracy: 0.9600\n",
            "Epoch 199/200\n",
            "2/2 - 11s - loss: 0.0378 - accuracy: 0.9573 - val_loss: 0.0506 - val_accuracy: 0.9400\n",
            "Epoch 200/200\n",
            "2/2 - 11s - loss: 0.0452 - accuracy: 0.9467 - val_loss: 0.0455 - val_accuracy: 0.9400\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f701463aa10>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 49
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aIa8snVGQZPR"
      },
      "source": [
        "result = model.evaluate(np.array(data_test), np.array(target_test))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cGly_vuJQayA"
      },
      "source": [
        "for name, value in zip(model.metrics_names, result):\n",
        "    print(name, value)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bcPy5U9OQeff"
      },
      "source": [
        "# plt.plot(model.history.history['accuracy'])\n",
        "# plt.plot(model.history.history['val_accuracy'])\n",
        "# plt.title('model accuracy')\n",
        "# plt.ylabel('accuracy')\n",
        "# plt.xlabel('epoch')\n",
        "# plt.legend(['train', 'validation'], loc='upper left')\n",
        "# plt.savefig('destination_path.eps', format='eps', dpi=1000)\n",
        "# plt.show()\n",
        "\n",
        "# # # summarize history for loss\n",
        "# # plt.plot(model.history['loss'])\n",
        "# # plt.plot(model.history['val_loss'])\n",
        "# # plt.title('model loss')\n",
        "# # plt.ylabel('loss')\n",
        "# # plt.xlabel('epoch')\n",
        "# # plt.legend(['train', 'validation'], loc='upper left')\n",
        "# # plt.savefig('destination_path1.eps', format='eps', dpi=1000)\n",
        "# # plt.show()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "W3dkSNeE2QLO"
      },
      "source": [
        "# saving the model \n",
        "\n",
        "import pickle \n",
        "# pickle_out = open(\"drive/MyDrive/CMP-258 Project/violence_det.pkl\", mode = \"wb\") \n",
        "# pickle.dump(history, pickle_out) \n",
        "# pickle_out.close()\n",
        "\n",
        "pickle.dump(model, open(\"/content/drive/MyDrive/CMP-258 Project/violence_det.pkl\", 'wb'))\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OYZFX1XrqT7p"
      },
      "source": [
        "import os\n",
        "import absl\n",
        "import datetime\n",
        "import tensorflow as tf\n",
        "import tensorflow_transform as tft\n",
        "\n",
        "from tfx.components.trainer.executor import TrainerFnArgs\n",
        "from tfx.components.trainer.fn_args_utils import DataAccessor\n",
        "from tfx_bsl.tfxio import dataset_options\n",
        "\n",
        "from keras.applications.inception_v3 import InceptionV3\n",
        "from keras.optimizers import SGD\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "from keras.models import Model\n",
        "from keras.layers import Dense, GlobalAveragePooling2D\n",
        "from keras.callbacks import ModelCheckpoint, TensorBoard, EarlyStopping\n",
        "\n",
        "\n",
        "def _input_fn(file_pattern: List[Text],\n",
        "              data_accessor: DataAccessor,\n",
        "              tf_transform_output: tft.TFTransformOutput,\n",
        "              batch_size: int = 200) -> tf.data.Dataset:\n",
        "\n",
        "  return data_accessor.tf_dataset_factory(\n",
        "      file_pattern,\n",
        "      dataset_options.TensorFlowDatasetOptions(\n",
        "          batch_size=batch_size, label_key=_transformed_name(_LABEL_KEY)),\n",
        "      tf_transform_output.transformed_metadata.schema)\n",
        "\n",
        "\n",
        "def _build_model(weights='imagenet'):\n",
        "    # create the base pre-trained model\n",
        "    base_model = InceptionV3(weights=weights, include_top=False)\n",
        "\n",
        "    # add a global spatial average pooling layer\n",
        "    x = base_model.output\n",
        "    x = GlobalAveragePooling2D()(x)\n",
        "    # let's add a fully-connected layer\n",
        "    x = Dense(1024, activation='relu')(x)\n",
        "    # and a logistic layer\n",
        "    predictions = Dense(2, activation='softmax')(x)\n",
        "\n",
        "    # this is the model we will train\n",
        "    model = Model(inputs=base_model.input, outputs=predictions)\n",
        "\n",
        "    for layer in model.layers[:-2]:\n",
        "        layer.trainable = False\n",
        "\n",
        "    # compile the model (should be done *after* setting layers to non-trainable)\n",
        "    model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['accuracy'])\n",
        "\n",
        "    return model\n",
        "\n",
        "\n",
        "\n",
        "def run_fn(fn_args: TrainerFnArgs):\n",
        "\n",
        "  tf_transform_output = tft.TFTransformOutput(fn_args.transform_output)\n",
        "\n",
        "  train_dataset = _input_fn(fn_args.train_files, fn_args.data_accessor, \n",
        "                            tf_transform_output, 40)\n",
        "  eval_dataset = _input_fn(fn_args.eval_files, fn_args.data_accessor, \n",
        "                           tf_transform_output, 40)\n",
        "  \n",
        "  model = _build_model()\n",
        "\n",
        "  tensorboard_callback = tf.keras.callbacks.TensorBoard(\n",
        "      log_dir=fn_args.model_run_dir, update_freq='batch')\n",
        "  model.fit(\n",
        "      train_dataset,\n",
        "      steps_per_epoch=fn_args.train_steps,\n",
        "      validation_data=eval_dataset,\n",
        "      validation_steps=fn_args.eval_steps,\n",
        "      callbacks=[tensorboard_callback])\n",
        "  \n",
        "  signatures = {\n",
        "      'serving_default':\n",
        "          _get_serve_tf_examples_fn(model,\n",
        "                                    tf_transform_output).get_concrete_function(\n",
        "                                        tf.TensorSpec(\n",
        "                                            shape=[None],\n",
        "                                            dtype=tf.string,\n",
        "                                            name='examples')),\n",
        "  }\n",
        "  model.save(fn_args.serving_model_dir, save_format='tf', signatures=signatures\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JbEwG1KXqUVw"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}